Hi all: I expect this is slightly off list but I am perplexed by a problem with our network which seems beyond the understanding of the suppliers. We have a brand new set of swiches with a gigabit backbone as a result of a leak following some dodgy building of the new server room. The switches are all D-Link. There are 8 24 port switches which are stacked in blocks of 4 and a master with the gig modules. Once we got the network running they kept falling over and resetting themselves to the factory default setting for the IP. The supplier and D-Link "solved" it by replacing all of them and they have worked for a few days but they have reset again. My question is, what would cause a switch to reset? It wouldn't reset if it was bombarded with packages? Would it be a bad NIC card somewhere in the system sending higher than normal voltage packages? All I can think of is that the power leads are causing problems. The suppliers have run the power cables through all of the data runs. Would this cause enough extra inteference to cause a switch to reset itself? Also (apologies to Thomas following his last missive about read the f** archives) but I can't find some discussion I had about squid settings sometime back. I believe it was mainly with Mark Evans but anyone else's oppinion would be welcome. I am setting up a box with the swgfl and they are unlocking port 80 for me. Would it be better to use 2 NICs or would just one using ipchains be enough to protect the internal network? I would appreciate anyones sample squid.conf with a 2 card setup to save me some time on books and how-tos. I realise this is rather lazy but I am a teacher and time is not a luxury I possess. Paul
Hi all:
I expect this is slightly off list but I am perplexed by a problem with our network which seems beyond the understanding of the suppliers. We have a brand new set of swiches with a gigabit backbone as a result of a leak following some dodgy building of the new server room. The switches are all D-Link. There are 8 24 port switches which are stacked in blocks of 4 and a master with the gig modules. Once we got the network running they kept falling over and resetting themselves to the factory default setting for the IP. The supplier and D-Link "solved" it by replacing all of them and they have worked for a few days but they have reset again. My question is, what would cause a switch to reset? It wouldn't reset if it was bombarded with packages? Would it be a bad NIC card somewhere in the system sending higher than normal voltage packages? All I can think of is that the power leads are causing problems. The suppliers have run the
Hi Paul,
This does sound strange!
I know that d-link are cheap but... (I'd recommend Lantech MaxSwitches -
easy after the event.)
Are the units doing a full reset. i.e they go through a POST or are they
resetting the tables only?
Do all of the units reset at the same time?
I assume that you are happy with the mains supply. Have you tried fitting a
UPS to filler out staffroom kettles etc. :-)
Another big problem in school is the need to have everything up at the same
time, but have you tried connecting one segment at a time.
Sorry, but all obvious stuff really :-(
Adrian
----- Original Message -----
From: Paul Taylor
cables through all of the data runs. Would this cause enough extra inteference to cause a switch to reset itself?
Just one other thought - Heat?
Later
Adrian
----- Original Message -----
From: Paul Taylor
Hi all:
I expect this is slightly off list but I am perplexed by a problem with our network which seems beyond the understanding of the suppliers. We have a brand new set of swiches with a gigabit backbone as a result of a leak following some dodgy building of the new server room. The switches are all D-Link.
On Thu, Oct 09, 2003 at 07:23:08PM +0100, Paul Taylor wrote:
Hi all:
I expect this is slightly off list but I am perplexed by a problem with our network which seems beyond the understanding of the suppliers. We have a brand new set of swiches with a gigabit backbone as a result of a leak following some dodgy building of the new server room. The switches are all D-Link. There are 8 24 port switches which are stacked in blocks of 4 and a master with the gig modules. Once we got the network running they kept falling over and resetting themselves to the factory default setting for the IP. The supplier and D-Link "solved" it by replacing all of them and they have worked for a few days but they have reset again. My question is, what would cause a switch to reset? It wouldn't reset if it was bombarded with packages? Would it be a bad NIC card somewhere in the system sending higher than normal voltage packages? All I can think of is
Should that reset the whole stack or just the module with that NIC plugged into? It might help to know if they found any faults on the units they replaced. Also do you have a note of the serial numbers of all of the units? It case they all come from the same production run.
that the power leads are causing problems. The suppliers have run the power cables through all of the data runs. Would this cause enough extra inteference to cause a switch to reset itself?
If these power cables are 230VAC, or even 415VAC then they don't know what they are doing. It is not acceptable to run mains cables with data cables. Telephone, alarm, CATV or PoE is generally ok to run with network cables so long as any voltages anywhere do not excede 120VDC.
Also (apologies to Thomas following his last missive about read the f** archives) but I can't find some discussion I had about squid settings sometime back. I believe it was mainly with Mark Evans but anyone else's oppinion would be welcome.
The SWGFL settings are something like. cache_peer proxy.st-peters-high.devon.sch.uk parent 8080 7 no-digest no-query default acl localservers dst 10.3.48.0/255.255.252.0 10.64.204.0/255.255.255.0 never_direct deny localservers never_direct allow all always_direct allow localservers always_direct deny all
I am setting up a box with the swgfl and they are unlocking port 80 for me. Would it be better to use 2 NICs or would just one using ipchains be enough to protect the internal network? I would appreciate anyones sample squid.conf with a 2 card setup to save me some time on books and how-tos. I realise this is rather lazy but I am a teacher and time is not a luxury I possess.
-- Mark Evans St. Peter's CofE High School Phone: +44 1392 204764 X109 Fax: +44 1392 204763
Hi Paul,
I missed the bit about the mains & data sharing the same conduit. This is
against regs. and you should get the installer to make good. There is a
recommended minimum distance for mains & data running in a parallel course,
even in separate conduits and they should not run in the same conduit. Tis
Friday and I cant remember the details. Note that 3 part trunking should be
screened, but little that is installed today is because users get away with
it working!
Same thing goes for avoiding fluorescent lamps, give a wide birth - I think
that's 0.5m.. Crossing mains at 90deg is okay but looks bad (hey, lets keep
those walls neat!)
The problem is that there's a lot of 'professional installers' out there
that would be better suited to house bashing and have learned their trade
from a mate, having never read the CAT5 regs let alone the electrical regs.
If you have induced voltages on the cat5 then the UPS will not help, you
could try surge protectors on the CAT5 but you'd do better to get the
cabling sorted first. If the manufactures send a rep to your site they'll be
looking for any 'let-out' rather than admit fault.
As Mark says, it's okay to mix lo voltages in the same conduit. including
telephones that can have 120V on the ringer! although 50V is more common on
conventional switches - rambling now!
Not sure how much this helps
Have a good weekend
Adrian
----- Original Message -----
From: Paul Taylor
===== Original Message From "adrian.wells"
===== Hi Paul, This does sound strange!
I know that d-link are cheap but... (I'd recommend Lantech MaxSwitches - easy after the event.)
Are the units doing a full reset. i.e they go through a POST or are they resetting the tables only?
I think it is the power cables as I just had a look (see the message to Mark) and tere are power cables all over the data cables.
Do all of the units reset at the same time?
I assume that you are happy with the mains supply. Have you tried fitting a UPS to filler out staffroom kettles etc. :-)
There are no UPS at present which could well solve a lot of it.
Another big problem in school is the need to have everything up at the same time, but have you tried connecting one segment at a time.
Sorry, but all obvious stuff really :-(
Adrian
Thanks Adrian,
participants (3)
-
adrian.wells
-
Mark Evans
-
Paul Taylor