We moved away from fibre a long time ago (it wasn't cost effective for our needs), but at the time we were using Qlogic HBAs and HP switches to connect to HP storage arrays and Storagetek tape drives. I don't recall having to use any proprietary drivers.

ISTR something about licenses for the fibre switch, as well as proprietary firmware for the storage arrays.

Sounds promising. And yes there are different licenses for the switch, which enable 4/8 or 16 ports of device. This device has an active licenses for all 16 ports. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On Monday 2020-02-24 14:49, Per Jessen wrote:

Date: Mon, 24 Feb 2020 14:49:33 From: Per Jessen To: opensuse@opensuse.org Subject: Re: [opensuse] switched fabric using qla2xxx pci card (fwd)

Paul Neuwirth wrote:

...

...

We moved away from fibre a long time ago (it wasn't cost effective for our needs), but at the time we were using Qlogic HBAs and HP switches to connect to HP storage arrays and Storagetek tape drives. I don't recall having to use any proprietary drivers.

ISTR something about licenses for the fibre switch, as well as proprietary firmware for the storage arrays.

Sounds promising. And yes there are different licenses for the switch, which enable 4/8 or 16 ports of device. This device has an active licenses for all 16 ports.

Overall I don't recall having any particular problems or difficulties. The configuration of the storage arrays was a little exotic, and we didn't have to touch the switches very much. AFAIR, the devices configured on the arrays would simply appear as regular scsi devices. Yes, the switches were also re-badged Brocades.

Good luck with the project, I guess 2nd hand 4Gb fibre equipment is quite affordable now?

The switch and the disk array where both between USD 100 and 150. Those Controllers no more than 20 each. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

24.02.2020 16:36, Andrei Borzenkov пишет:

24.02.2020 16:10, Per Jessen пишет:

Somehow I did not receive original post.

...

Paul Neuwirth wrote:

...

Hello group,

I am trying to evaluate use of fibre channel switched fabric. But I am not able to get it working (openSUSE Leap 15.1).

That's not enough to even start guessing. Explain what you did, what you expected and what you got.

...

I would appreciate,

...

if anybody knows something or where to find (working) instructions. All instructions I found refer to old/or proprietary drivers/methods (kernel 2.6 or older).

I have no idea what you are talking about. FibreChannel normally just works. The only potential setting is protocol (FC-AL or point-to-point) and HBA usually tries both so it just works.

You need to make sure you are using the same type of SFP and cables (MMF vs. SMF) everywhere.

...

...

I use following hardware: -HP branded (AE312 - 60001) QLogic Corp. ISP2432 4G HBA PCIe

Should just work.

...

...

-IBM branded (2005-B16) Brocade 200E 16 Port FC switch

That should be 4G so compatible with 4G HBA. You would need to either remove zoning completely (so each device see each device) or create zones for your HBA and storage devices.

...

...

-DS14MK2 Disk Shelf

That's not a storage. It is disk shelf for NetApp FAS (very old models) and was never intended to be used as direct JBOD. This only supports FC-AL protocol and I do not know whether it supports fabric attachment (it has no reason to support it). So the first thing to test would be to connect it directly, without switch. If HBA sees disks and you insist on using switch, Brocade supports QuickLoop which simulates FC-AL between different ports. It may work.

Somehow I still do not receive messages from OP, but looking in archives 16 0 id N2 Online L-Port 9 public, 7 private 17 1 id N2 Online L-Port 6 public, 10 private this confirms what I wrote. These two ports cannot talk with other ports on switch. Try direct connection first.

...

...

All of those products are EOS and I am not able to get proprietary drivers or firmware updates.

HBA should work with upstream driver, FC switch usually just works as well (there is very little reasons to upgrade unless you hit specific bug or need additional features). DS14mk2 could be an issue though. Also native NetApp disks are 520bps which may or may not be a problem.

Anyway, start with explaining what exactly you did and what "does not work" currently.

-- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On Monday 2020-02-24 16:28, Dave Howorth wrote:

Date: Mon, 24 Feb 2020 16:28:31 From: Dave Howorth To: opensuse@opensuse.org Subject: missing messages was: Re: [opensuse] switched fabric using qla2xxx pci card (fwd)

On Mon, 24 Feb 2020 16:42:12 +0300 Andrei Borzenkov wrote:

...

Somehow I did not receive original post.

...

Somehow I still do not receive messages from OP, but looking in archives

FWIW, I didn't see the original message either. I did see Paul's second post but not his third nor yet his fourth.

I haven't received any relevant bounce warnings either.

I noticed that issue a while ago, when I set up SPF and DMARC on the mail server. Back then I didn't find a solution quickly and postponed it. Since message get's altered but DMARC headers are left in tact the validation fails. And due to mail list server sending with original from, SPF will fail too. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On Monday 2020-02-24 21:05, Dave Howorth wrote:

Date: Mon, 24 Feb 2020 21:05:32 From: Dave Howorth To: opensuse@opensuse.org Subject: [opensuse] Re: missing messages

On Mon, 24 Feb 2020 16:51:07 +0100 (CET) Paul Neuwirth wrote:

...

On Monday 2020-02-24 16:28, Dave Howorth wrote:

...

Date: Mon, 24 Feb 2020 16:28:31 From: Dave Howorth To: opensuse@opensuse.org Subject: missing messages was: Re: [opensuse] switched fabric using qla2xxx pci card (fwd)

On Mon, 24 Feb 2020 16:42:12 +0300 Andrei Borzenkov wrote:

...

Somehow I did not receive original post.

...

Somehow I still do not receive messages from OP, but looking in archives

FWIW, I didn't see the original message either. I did see Paul's second post but not his third nor yet his fourth.

I haven't received any relevant bounce warnings either.

I noticed that issue a while ago, when I set up SPF and DMARC on the mail server. Back then I didn't find a solution quickly and postponed it. Since message get's altered but DMARC headers are left in tact the validation fails. And due to mail list server sending with original from, SPF will fail too.

That sounds like a bug in the list mailer's configuration, no? (correction of what I wrote: DKIM (not DMARC) fails on changed content.

Not quite sure. The only "safe" way I know is changing the senders mail address. e.g. something like originaluser+originaldomain@opensuse.org and the mail server forwarding to originaluser@originaldomain. at least that is what I would try to set up. other not that failsafe way I found would be adding "from_is_list" header and removing DKIM headers. It is a real problem for me on the opensuse lists, why I havent been active at all since this mailserver update. I introduced that because of massive spam, sent with adresses using my domains - but in fact not originating. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On Mon, 24 Feb 2020 14:50:36 -0800 L A Walsh wrote:

I think you are hitting a feature by gmail.

None of us use gmail, so I don't think so. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

* Patrick Shanahan [02-25-20 08:40]:

* Dave Howorth [02-25-20 05:42]:

...

On Mon, 24 Feb 2020 14:50:36 -0800 L A Walsh wrote:

...

I think you are hitting a feature by gmail.

None of us use gmail, so I don't think so.

fwiw: yesterday I received a notice from opensuse+owner saying "Bouncing messages from opensuse@opensuse.org" and there were six listed and four were from the OP, probably the referenced posts.

fwiw2: I regularly receive these notices from opensuse-***+owner and they usually reference posts from the same small group of people. Guessing these people have some small irregularity or feature disliked by the opensuse list software and it chokes. Of course it might be my setup causing the bounces but, ... -- (paka)Patrick Shanahan Plainfield, Indiana, USA @ptilopteri http://en.opensuse.org openSUSE Community Member facebook/ptilopteri Photos: http://wahoo.no-ip.org/piwigo paka @ IRCnet freenode -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On Tue, 25 Feb 2020 15:31:37 +0100 Per Jessen wrote:

Patrick Shanahan wrote:

...

* Patrick Shanahan [02-25-20 08:40]:

...

* Dave Howorth [02-25-20 05:42]:

...

On Mon, 24 Feb 2020 14:50:36 -0800 L A Walsh wrote:

...

I think you are hitting a feature by gmail.

None of us use gmail, so I don't think so.

fwiw: yesterday I received a notice from opensuse+owner saying "Bouncing messages from opensuse@opensuse.org" and there were six listed and four were from the OP, probably the referenced posts.

fwiw2: I regularly receive these notices from opensuse-***+owner and they usually reference posts from the same small group of people. Guessing these people have some small irregularity or feature disliked by the opensuse list software and it chokes. Of course it might be my setup causing the bounces but, ...

When you receive that listing, it is because your receiving mailserver has bounced something the list server tried to deliver.

This is one of the bounce messages:

554 5.7.9 Message not accepted for policy reasons.

One example is Paul who has an SPF record that prohibits mails sent by his domain from any other source than his MX.

I'm not sure I understand. You mean Paul sends a message that contains an SPF record (which I've heard of but don't really understand) and that the list forwards his mail to list recipients still containing the same SPF record? And some recipient's mailservers (presumably my ISP?) who bother to check for such things consequently discard it as from the wrong source? If so, surely the list server should remove the SPF record before forwarding the message, and perhaps add one of its own? -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On Wed, 26 Feb 2020 08:13:42 +0100 Bengt Gördén wrote:

On 2020-02-25 21:20, Dave Howorth wrote:

...

And some recipient's mailservers (presumably my ISP?) who bother to check for such things consequently discard it as from the wrong source?

That's the whole point of Sender Policy Framework (spf)

...

If so, surely the list server should remove the SPF record before forwarding the message, and perhaps add one of its own?

You got it all wrong. It's not included in the mail. It's part of DNS, and the mail server software (sendmail, postfix etc.) looks at it.

If you look at he's (Pauls) mx domain you get mail.swabian.net.

A check of the spf record gives:

$ dig +short mail.swabian.net. TXT "v=spf1 a -all"

The records are interpreted from left to right.

v = version a = allow sending mail from the DNS a-record -all = reject all mail coming from hosts other than the a-record.

$ dig +short mail.swabian.net. 80.152.201.148

A starting point to read is Wikipedia. https://en.wikipedia.org/wiki/Sender_Policy_Framework

Thanks for the explanation. As I said I don't understand SPF. The wikipedia article just looks like gobbledygook to me. So is there anything that either Paul, or the list mailserver, or for that matter my ISP's mailserver is doing 'wrong' somehow that could be improved to prevent Paul's mails getting bounced? Or is it just a poorly designed protection scheme? -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

Carlos E. R. wrote:

Basically Paul mail setup is configured on his DNS (so, no headers to delete) to tell others (like your ISP) to reject any mail claiming to come from him (From header), that doesn't actually get posted by his IP. As list mail is coming from SUSE IP, it gets correctly rejected. So, he either disables SPF, or uses a different mail address.

Well, there is more than one way to skin a cat :-) a) disable SPF. Paul mentioned he'd had trouble with his address being abused for spamming, so he probably does not want to do that. b) update the SPF record to permit mails being sent from mx[12].suse.de. This is a kludge. c) Paul could relax his SPF policy to SOFTFAIL (~all) d) Ask Yahoo to whitelist mx[12].suse.de. e) SUSE could implement SRS on the main mailserver f) openSUSE could move to our own MX'es and implement SRS. For now, the easiest solution is no doubt (c). The best options are (d), (e) or (f). The first half of (f) is in progress. -- Per Jessen, Zürich (3.4°C) Member, openSUSE Heroes -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

Not quite, no. :-) This is interesting.  I just got an education on something that I never heard of.  Thanks. I like and appreciate simple summary descriptions

On 2/26/2020 4:19 AM, Per Jessen wrote: that are easier to understand than diving into the fine manual.

In his DNS, Paul has set up an SPF record for his domain "paul-neuwirth.nl". This record is public and informs everyone which server is allowed to send mails from "@paul-neu...". It also Do I understand correctly that one solution would be for a person using SPF would be to add mx.suse.de to his/her SPF list so this doesn't happen?

Damon Register -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On Wednesday 2020-02-26 12:38, Per Jessen wrote:

Date: Wed, 26 Feb 2020 12:38:58 From: Per Jessen To: opensuse@opensuse.org Subject: Re: EXTERNAL: Re: [opensuse] Re: missing messages -- SPF violations

Damon Register wrote:

...

On 2/26/2020 4:19 AM, Per Jessen wrote:

...

Not quite, no. :-) This is interesting.  I just got an education on something that I never heard of.  Thanks. I like and appreciate simple summary descriptions that are easier to understand than diving into the fine manual.

...

In his DNS, Paul has set up an SPF record for his domain "paul-neuwirth.nl". This record is public and informs everyone which server is allowed to send mails from "@paul-neu...". It also

Do I understand correctly that one solution would be for a person using SPF would be to add mx.suse.de to his/her SPF list so this doesn't happen?

Actually the list mails are sent out via two other machines :

proxy-nue1.opensuse.org proxy-nue2.opensuse.org

I have not thought it through, but adding those two to an SPF record would probably work, yes. I wouldn't recommend it though - our list server is far from the only one with this problem, that SPF record could end up being quite long :-)

was absent a few days. I did change accordingly, added those servers to SPF records (and hopefully did not mess anything up). But I assume DKIM will still fail.

On Friday 2020-02-28 16:13, Per Jessen wrote:

Date: Fri, 28 Feb 2020 16:13:00 From: Per Jessen To: opensuse@opensuse.org Subject: Re: EXTERNAL: Re: [opensuse] Re: missing messages -- SPF violations

Paul Neuwirth wrote:

...

On Wednesday 2020-02-26 12:38, Per Jessen wrote:

...

Date: Wed, 26 Feb 2020 12:38:58 From: Per Jessen

Actually the list mails are sent out via two other machines :

proxy-nue1.opensuse.org proxy-nue2.opensuse.org

I have not thought it through, but adding those two to an SPF record would probably work, yes. I wouldn't recommend it though - our list server is far from the only one with this problem, that SPF record could end up being quite long :-)

was absent a few days. I did change accordingly, added those servers to SPF records (and hopefully did not mess anything up).

What I see right now won't work - you include '_sfp.swabian.net', but that loops back to www.swabian.net, which again say 'include _sfp.swabian.net'.

.. TTL is 7200 of the _sfp. records, I only waited 2 hours to post, so I assume the result was outdated.

...

But I assume DKIM will still fail.

DKIM really ought to work, but I guess one problem is that we modify the Subject line.

Rather I think it is the bottom text. or both if subject line is also included in the hash checks.

-- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

-- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On Friday 2020-02-28 18:53, Per Jessen wrote:

Date: Fri, 28 Feb 2020 18:53:19 From: Per Jessen To: opensuse@opensuse.org Subject: Re: EXTERNAL: Re: [opensuse] Re: missing messages -- SPF violations

Hmm, I still see the same:

# dig paul-neuwirth.nl txt +short "v=spf1 include:_sfp.swabian.net -all" "google-site-verification=PAoK2X1VM82ESkMkpdvwzsjyexhK9olhjWhCKGb1DGk"

TTL = 1800

# dig _sfp.swabian.net txt +short www.swabian.net. "v=spf1 include:_spf.swabian.net -all"

TTL = 3600.

ty for the checking. I just noticed some mistyping.. _sfp instead of _spf, thus you get a CNAME record (for the non existant _sfp). corrected, but need to wait again. interestingly, I did not get any (dmarc) reports in last time with that misconfig. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

L A Walsh wrote:

Seems like when the email is resent from suse, suse should add it's own SPF --

That's not how it works though. It cnanot be "added" as part of processing. suse.com does publish an SPF record, but opensuse.org does not, for good reason. (it would play havoc with our member email addresses).

but enforcing the SPF rules against forwarding servers (whether internal to an email list or internal to a receiving ISP) seems wrong --

Yes, and many ISPs do whitelist mailing list servers.

like SPF was never intended to be compared against 'in-transit' hosts -- only the source of the original message, no?

Yup.

Certainly with SPF, it should only be enforced against the original sender.

BTW, where does DKIM fit into all this?

It is a method to detect forged email addresses. With DKIM, you sign your outgoing emails with your domain key. Receivers can verify the signature with a simple DNS lookup. -- Per Jessen, Zürich (8.2°C) http://www.cloudsuisse.com/ - your owncloud, hosted in Switzerland. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On Wed, 26 Feb 2020 08:07:03 -0500 Carl Hartung wrote:

On Wed, 26 Feb 2020 10:19:40 +0100 Per Jessen wrote:

8< - - - - - snipped - - - - - >8

...

d) sent to a mailserver, in this case one of Yahoo's.

e) Yahoo does an SPF check, which fails as the mail from @paul-neu... is now being sent by mx.suse.de which is not listed in the SPF record.

f) the policy says to refuse any mail not coming from a listed server, so Yahoo correctly / immediately bounces the mail.

g) the bounce eventually ends up on the list server being flagged as "could not deliver to paka@".

Hi All,

Yahoo Mail is notoriously hypersensitive, and, it's users can only declare specific senders (i.e. the "From:" address) or domains as 'safe' or 'approved.'

In this case, if the Yahoo Mail user declares "opensuse.org" as 'safe' or 'approved,' barring any other as yet undisclosed factors, this would most likely mitigate the problem.

But the problem is not just with Yahoo, and indeed it doesn't seem like Yahoo are doing anything wrong. Plus why should all (non) recipients of these messages have to do special things?

regards,

Carl

-- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On Wed, 26 Feb 2020 16:08:28 +0100 Per Jessen wrote:

Dave Howorth wrote:

...

On Wed, 26 Feb 2020 08:07:03 -0500 Carl Hartung wrote:

...

Yahoo Mail is notoriously hypersensitive, and, it's users can only declare specific senders (i.e. the "From:" address) or domains as 'safe' or 'approved.'

In this case, if the Yahoo Mail user declares "opensuse.org" as 'safe' or 'approved,' barring any other as yet undisclosed factors, this would most likely mitigate the problem.

But the problem is not just with Yahoo, and indeed it doesn't seem like Yahoo are doing anything wrong.

They are not, they are acting absolutely correct. It would be nice if they could whitelist our mailing list server though. (

But it seems extremely unreasonable for every 'absolutely correct' ISP in the world to have to list every legitimate mail server in the world, manually! -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On Wed, 26 Feb 2020 17:09:36 +0100 Per Jessen wrote:

Dave Howorth wrote:

...

On Wed, 26 Feb 2020 16:08:28 +0100 Per Jessen wrote:

...

Dave Howorth wrote:

...

On Wed, 26 Feb 2020 08:07:03 -0500 Carl Hartung wrote:

...

Yahoo Mail is notoriously hypersensitive, and, it's users can only declare specific senders (i.e. the "From:" address) or domains as 'safe' or 'approved.'

In this case, if the Yahoo Mail user declares "opensuse.org" as 'safe' or 'approved,' barring any other as yet undisclosed factors, this would most likely mitigate the problem.

But the problem is not just with Yahoo, and indeed it doesn't seem like Yahoo are doing anything wrong.

They are not, they are acting absolutely correct. It would be nice if they could whitelist our mailing list server though. (

But it seems extremely unreasonable for every 'absolutely correct' ISP in the world to have to list every legitimate mail server in the world, manually!

s/mail server/mailing list server/

Oops, sorry.

It is the easiest way out, unless the mailing list server implements SRS rewriting which will also solve the problem.

Well mailing list server implementing SRS rewriting seems like a much better solution. Order N instead of order NxM (where I expect M is much larger than N) for starters. Plus it's the mail list servers that are actually causing the problem, if you discount the existence of SPF as being the problem. I'm not sure I entirely understand the wikipedia article on SRS either. Does it imply that I would receive messages from the list with weird from addresses or is that addressed somehow? Most of the mailing lists I'm a member of show the originator's name as 'from'. Does that mean they don't use SRS or that something is munging what it displays to me? openSUSE is the only mail list where I get occasional bounce advices, which I've never understood up until now. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

* Per Jessen [02-26-20 10:09]:

Dave Howorth wrote:

...

On Wed, 26 Feb 2020 08:07:03 -0500 Carl Hartung wrote:

...

Yahoo Mail is notoriously hypersensitive, and, it's users can only declare specific senders (i.e. the "From:" address) or domains as 'safe' or 'approved.'

In this case, if the Yahoo Mail user declares "opensuse.org" as 'safe' or 'approved,' barring any other as yet undisclosed factors, this would most likely mitigate the problem.

But the problem is not just with Yahoo, and indeed it doesn't seem like Yahoo are doing anything wrong.

They are not, they are acting absolutely correct. It would be nice if they could whitelist our mailing list server though. (

and to make it more complicated, yahoo filters are quite specific only allowing "From", "To/CC", "Subject" and "Body" so spf/header/dkim are not available. And yahoo must believe all gmail/googlemail addresses are spammers or they are just showing their corporate pissing post. -- (paka)Patrick Shanahan Plainfield, Indiana, USA @ptilopteri http://en.opensuse.org openSUSE Community Member facebook/ptilopteri Photos: http://wahoo.no-ip.org/piwigo paka @ IRCnet freenode -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

Patrick Shanahan wrote:

but still somewhat odd that some of the bounced mail authors are delivered "sometimes" and not others. As Paul's mails are in subject, I do get some he posts and some are bounced.

Any chance these were direct replies to you? Any direct replies would go through just fine, I think. It is also not 100% sure that Yahoo rejects due to SPF - there _could_ be other factors, but the reject message doesn't say. -- Per Jessen, Zürich (8.7°C) Member, openSUSE Heroes. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

Somehow I did not receive original post.

I believe it is because of DMARC setup - got some DMARC reports. That'll require another topic or some research.

...

Paul Neuwirth wrote:

...

Hello group,

I am trying to evaluate use of fibre channel switched fabric. But I am not able to get it working (openSUSE Leap 15.1).

That's not enough to even start guessing. Explain what you did, what you expected and what you got.

I connected a Server running openSUSE Leap 15.1 with a FC HBA built-in to a switch connected also to a FC Disk Shelf. Using AVAGO 850nm SFPs and orange LC-LC optical cables if I remember correct like that: HBA Port 1 2 | | Switch Port 1 2 3 4 | | Shelf Contr. 1 IN OUT Contr 2 IN OUT (do not know if it was in the jack named "IN" or "OUT", but I do remember, only one version worked, i.e. the devices and the hard drives showed up on the switch. Controller was set to default values: ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ +++++/------------------Selected Adapter-------------------------\++++ +++++| Adapter Type Address Slot Bus Device Function |++++ +++++| HPAE312A 4000 01 01 00 1 |++++ +++++\-----------------------------------------------------------/++++ ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ ++++++/-----------------Adapter Settings--------------------\+++++++++ ++++++| |+++++++++ ++++++| BIOS Add/----------------------------------------------------\ ++++++| BIOS Rev| Option Type of Connection | ++++++| Adapter | | ++++++| Interrup| 0 - Loop only | ++++++| Adapter | 1 - Point to point only | ++++++| Host Ada| 2 - Loop preferred, otherwise point to point | ++++++| Frame Si\----------------------------------------------------/ ++++++| Loop Reset Delay: 5 |+++++++++ ++++++| Adapter Hard Loop ID: Disabled |+++++++++ ++++++| Hard Loop ID: 0 |+++++++++ ++++++| Connection Options: 2 |+++++++++ ++++++| Fibre Channel Tape Support:Enabled |+++++++++ ++++++| Data Rate: 2 |+++++++++ ++++++| |+++++++++ + Use <Arrow-keys> and <Enter> to change settings, <Esc> to exit I expected, that those HDs can be accessed (or even found) on the server. Loopback data test using the Fast!UTIL works. also the WWNs of the HDs show up in dmesg. But I cannot access/find these devices in the OS or Fast!UTIL

...

I would appreciate,

...

if anybody knows something or where to find (working) instructions. All instructions I found refer to old/or proprietary drivers/methods (kernel 2.6 or older).

I have no idea what you are talking about. FibreChannel normally just works. The only potential setting is protocol (FC-AL or point-to-point) and HBA usually tries both so it just works.

You need to make sure you are using the same type of SFP and cables (MMF vs. SMF) everywhere.

...

...

I use following hardware: -HP branded (AE312 - 60001) QLogic Corp. ISP2432 4G HBA PCIe

Should just work.

...

...

-IBM branded (2005-B16) Brocade 200E 16 Port FC switch

That should be 4G so compatible with 4G HBA. You would need to either remove zoning completely (so each device see each device) or create zones for your HBA and storage devices.

tried previously without zoning and tried again with zoning disabled. no success

...

...

-DS14MK2 Disk Shelf

That's not a storage. It is disk shelf for NetApp FAS (very old models) and was never intended to be used as direct JBOD. This only supports FC-AL protocol and I do not know whether it supports fabric attachment (it has no reason to support it). So the first thing to test would be to connect it directly, without switch. If HBA sees disks and you insist on using switch, Brocade supports QuickLoop which simulates FC-AL between different ports. It may work.

I couldn't yet find out how to set up quick loop (using the GUI no devices show in the selection list). First I try the setup without switch.

...

...

All of those products are EOS and I am not able to get proprietary drivers or firmware updates.

HBA should work with upstream driver, FC switch usually just works as well (there is very little reasons to upgrade unless you hit specific bug or need additional features). DS14mk2 could be an issue though. Also native NetApp disks are 520bps which may or may not be a problem.

I know, but having built-in "normal" hard disks. These show up in the switch-ui. According to a source, I do not remember, it should just work plug and pray..

Anyway, start with explaining what exactly you did and what "does not work" currently.

Somehow I still do not receive messages from OP, but looking in archives 16 0 id N2 Online L-Port 9 public, 7 private 17 1 id N2 Online L-Port 6 public, 10 private this confirms what I wrote. These two ports cannot talk with other ports on switch. Try direct connection first.

that is what I thought the switch is (also good for), not to be restricted to the Loop-setup. Also according to the GUI that should work. IIRC, I also read, that someone did a Firmware Upgrade on th DS14mk2, which also enabled use of N-Ports (?). The controllers do have a console plug, but with PS/2 connectors(?). I didn't find out how to connect to RS-9 correctly. But, what you wrote, may be right, as that would explain everything. But HBA should support P2P and Loop. I tried direct connections: HBA Port 1 | Shelf Contr 1 IN OUT | Shelf Contr 2 IN OUT Fast!UTIL: loopback test works. scan devices doesn't show any. OS: nothing found (lsscsi / rescan-scsi-bus.sh) rmmod/modprobe: https://www.swabian.net/extern/opensuse/fcdmesg2.txt connected in another way: HBA Port 1 | Shelf Contr 1 IN OUT | Shelf Contr 2 IN OUT no change. https://www.swabian.net/extern/opensuse/fcdmesg3.txt Thank you for your support so far. I do not have experience with FC besides this case. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On Mon, Feb 24, 2020 at 7:01 PM Paul Neuwirth wrote:

I now changed cabling as follows:

HBA Port 1 2 | | Shelf Contr 1 IN OUT | | | Shelf Contr 2 IN OUT | |________/

now each hard disk is shown 4 times.

Unfortunately it is garbled but it seems you interconnected both IO modules. This is wrong. You are supposed to see exactly 2 paths to each disk. You should connect HBA ports to IN ports on IO modules. Do not interconnect IO modules themselves.

as each of the two ports are connected to both controllers.

# lsscsi [... local hdds on host 0 and 1 ...] [2:0:0:0] enclosu XYRATEX RS-1402-SA-XNS1 3034 - [2:0:1:0] enclosu XYRATEX RS-1402-SA-XNS1 3034 - [2:0:2:0] disk XYRATEX RS-1402-SA-XNSSX /dev/sdai [2:0:3:0] disk XYRATEX RS-1402-SA-XNSSX /dev/sdd [2:0:4:0] disk ST4000NM 0033-9ZM170 SX GA06 /dev/sde

Yes, that looks like SATA disk. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On Monday 2020-02-24 17:46, Andrei Borzenkov wrote:

Date: Mon, 24 Feb 2020 17:46:42 From: Andrei Borzenkov To: SuSE Linux Subject: Re: [opensuse] switched fabric using qla2xxx pci card (fwd)

On Mon, Feb 24, 2020 at 6:03 PM Paul Neuwirth wrote:

...

...

...

...

-DS14MK2 Disk Shelf

That's not a storage. It is disk shelf for NetApp FAS (very old models) and was never intended to be used as direct JBOD. This only supports FC-AL protocol and I do not know whether it supports fabric attachment (it has no reason to support it).

Correction - there are two variants of DS14mk2: DS14mk2-FC and DS14mk2-AT. The former is native FC and the latter is SATA disks with SATA-to-FC dongle. The DS14mk2-FC was supported in Fabric MetroCuster where shelves are connected to switch, so it should actually work. DX14mk2-AT was never support in FMC.

Judging by disk models you have DS14mk2-AT.

that's right it's the AT variant. These "SATA-to-FC dongles" are hopefully causing the next major problem. The 4TB disks are only accessible as 1.64TiB HDDs Now it's important to know, which part causes this. I hope that it is in fact the NetApp disk array or electronic of the disk bays. If it is, I have to find some other hardware. Anyone got tips (FC 4G, redundant storage controllers and redundant power supplies, 14+ SATA-drives)? thank you (parted) p Model: ST4000NM 0035-1V4107 SX (scsi) Disk /dev/sdh: 1.64TiB Sector size (logical/physical): 512B/512B Partition Table: gpt Disk Flags: P.S.: now it "works" also with the switch, maybe due to the change of mode of operation to loop only, switch shows now L-Port for the HBAs too:

switchshow switchName: fcsw1 switchType: 34.0 switchState: Online switchMode: Native switchRole: Principal switchDomain: 1 switchId: fffc01 switchWwn: 10:00:00:05:1e:03:d8:ae zoning: OFF switchBeacon: OFF

Area Port Media Speed State ============================== 16 0 id N4 Online L-Port 1 public 17 1 id N4 Online L-Port 1 public 18 2 id N2 Online L-Port 16 public 19 3 id N2 Online L-Port 16 public -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org