[opensuse-buildservice] obs 2.5.1 as local installation - a first glance - openSUSE Build Service

newer
[opensuse-buildservice] Open Build...

[opensuse-buildservice] obs 2.5.1 as local installation - a first glance

Stefan Botter

22 Apr 2014 22 Apr '14

08:47

Hi all, after updating my two instances (my university's build server and PMBS) here is my first impression: The update went more or less smooth. + with 2.5.1 the db update runs through without any error - thank you, Adrian! - README.UPDATERS still has the task to db:migrate the webui DB, which does not exist anymore (bee github #650) - the search in the web frontend does not work, thinking sphinx seems not being able to talk to the database - the system is catastrophically slow, and I mean "slow" with a capital "S"! Compared with the performance of 2.4.6 it seems to need the triple amount of CPU, and four times disk IO - these data come from PMBS. Is there someone, who has tried to downgrade to 2.4.6 again, and is this possible at all, apart from replaying a backup? Greetings, Stefan -- Stefan Botter Network Manager Jacobs University Bremen gGmbH Campus Ring 1 | 28759 Bremen | Germany Commercial registry: Amtsgericht Bremen, HRB 18117 CEO / Geschäftsführerin: Prof. Dr.-Ing. Katja Windt Chair Board of Governors: Prof. Dr. Karin Lochte

Attachments:

signature.asc (application/pgp-signature — 222 bytes)

Show replies by date

Adrian Schröter

22 Apr 22 Apr

09:14

On Dienstag, 22. April 2014, 10:47:56 wrote Stefan Botter:

...

Hi all,

after updating my two instances (my university's build server and PMBS) here is my first impression:

The update went more or less smooth. + with 2.5.1 the db update runs through without any error - thank you, Adrian! - README.UPDATERS still has the task to db:migrate the webui DB, which does not exist anymore (bee github #650)

dropped now.

...

- the search in the web frontend does not work, thinking sphinx seems not being able to talk to the database

If you have the config/production.sphinx.conf file, a simple rake ts:index ts:start should index and start the searchd . The original idea was that clockworkd is starting it ....

...

- the system is catastrophically slow, and I mean "slow" with a capital "S"! Compared with the performance of 2.4.6 it seems to need the triple amount of CPU, and four times disk IO - these data come from PMBS.

Do you speak about api, scheduler, worker or what ?

...

Is there someone, who has tried to downgrade to 2.4.6 again, and is this possible at all, apart from replaying a backup?

I doubt that the db rollback will work that far.... -- Adrian Schroeter email: adrian@suse.de SUSE LINUX GmbH, GF: Jeff Hawn, Jennifer Guild, Felix Imendörffer, HRB 21284 (AG Nürnberg) Maxfeldstraße 5 90409 Nürnberg Germany -- To unsubscribe, e-mail: opensuse-buildservice+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-buildservice+owner@opensuse.org

Stefan Botter

13:55

Hi Adrian, On Tue, 22 Apr 2014 11:14:21 +0200 "Adrian Schröter" wrote:

...

On Dienstag, 22. April 2014, 10:47:56 wrote Stefan Botter:

...
- the search in the web frontend does not work, thinking sphinx seems not being able to talk to the database

If you have the config/production.sphinx.conf file, a simple

rake ts:index ts:start

should index and start the searchd . The original idea was that clockworkd is starting it ....

Great! The search works now. Will it start automatically the next time I boot the machine and start up the services?

...

...
- the system is catastrophically slow, and I mean "slow" with a capital "S"! Compared with the performance of 2.4.6 it seems to need the triple amount of CPU, and four times disk IO - these data come from PMBS.

Do you speak about api, scheduler, worker or what ?

PMBS is running everything except a worker, so mysql, api, src_server, rep_server, signer, scheduler, dispatcher. Please see the attached images from pnp4nagios. I updated PMBS on Apr 19th, and around 3 pm the system started to burn ... The system is a 8GB, 4CPU VM in vCenter, on a OK-performing RAID10 of 14 disks.

...

...
Is there someone, who has tried to downgrade to 2.4.6 again, and is this possible at all, apart from replaying a backup?

I doubt that the db rollback will work that far....

Thought so, but I will secure my backup from 2.4.6 to be able to go back - for the time being. Another thing: where do I report setup issues with the appliance to? Github, or Bugzilla? Greetings, Stefan -- Stefan Botter Network Manager Jacobs University Bremen gGmbH Campus Ring 1 | 28759 Bremen | Germany Commercial registry: Amtsgericht Bremen, HRB 18117 CEO / Geschäftsführerin: Prof. Dr.-Ing. Katja Windt Chair Board of Governors: Prof. Dr. Karin Lochte

Adrian Schröter

14:02

On Dienstag, 22. April 2014, 15:55:49 wrote Stefan Botter:

...

Hi Adrian,

On Tue, 22 Apr 2014 11:14:21 +0200 "Adrian Schröter" wrote:

...
On Dienstag, 22. April 2014, 10:47:56 wrote Stefan Botter:

...
- the search in the web frontend does not work, thinking sphinx seems not being able to talk to the database

If you have the config/production.sphinx.conf file, a simple

rake ts:index ts:start

should index and start the searchd . The original idea was that clockworkd is starting it ....

Great! The search works now. Will it start automatically the next time I boot the machine and start up the services?

...
...
- the system is catastrophically slow, and I mean "slow" with a capital "S"! Compared with the performance of 2.4.6 it seems to need the triple amount of CPU, and four times disk IO - these data come from PMBS.

Do you speak about api, scheduler, worker or what ?

PMBS is running everything except a worker, so mysql, api, src_server, rep_server, signer, scheduler, dispatcher. Please see the attached images from pnp4nagios. I updated PMBS on Apr 19th, and around 3 pm the system started to burn ...

hm, a bit more detailed debug logs would be needed. If it is just the delayed jobs for example, it might be that just the data update/scanning jobs do have some more long running work directly after the update. But it will go away ... Can you check which processes do use the CPU/IO mostly? Maybe an strace on that might be helpful.

...

The system is a 8GB, 4CPU VM in vCenter, on a OK-performing RAID10 of 14 disks.

...
...
Is there someone, who has tried to downgrade to 2.4.6 again, and is this possible at all, apart from replaying a backup?

I doubt that the db rollback will work that far....

Thought so, but I will secure my backup from 2.4.6 to be able to go back - for the time being.

Another thing: where do I report setup issues with the appliance to? Github, or Bugzilla?

Greetings,

Stefan

github is fine. -- Adrian Schroeter email: adrian@suse.de SUSE LINUX GmbH, GF: Jeff Hawn, Jennifer Guild, Felix Imendörffer, HRB 21284 (AG Nürnberg) Maxfeldstraße 5 90409 Nürnberg Germany -- To unsubscribe, e-mail: opensuse-buildservice+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-buildservice+owner@opensuse.org

Stefan Botter

25 Apr 25 Apr

12:03

Hi Adrian, On Tue, 22 Apr 2014 16:02:13 +0200 "Adrian Schröter" wrote:

...

On Dienstag, 22. April 2014, 15:55:49 wrote Stefan Botter:

...
PMBS is running everything except a worker, so mysql, api, src_server, rep_server, signer, scheduler, dispatcher. Please see the attached images from pnp4nagios. I updated PMBS on Apr 19th, and around 3 pm the system started to burn ...

hm, a bit more detailed debug logs would be needed. If it is just the delayed jobs for example, it might be that just the data update/scanning jobs do have some more long running work directly after the update. But it will go away ...

Can you check which processes do use the CPU/IO mostly? Maybe an strace on that might be helpful.

The schedulers for i586 and x86_64 seem to burn a lot more CPU cycles than compared to 2.4.6, and this results in more I/O. If Henne is near you, he may help in having a more direct look at the machine, else please tell me, what I should look for. Greetings, Stefan -- Stefan Botter Network Manager Jacobs University Bremen gGmbH Campus Ring 1 | 28759 Bremen | Germany Commercial registry: Amtsgericht Bremen, HRB 18117 CEO / Geschäftsführerin: Prof. Dr.-Ing. Katja Windt Chair Board of Governors: Prof. Dr. Karin Lochte

3659

Age (days ago)

3662

Last active (days ago)

List overview

Download

4 comments

2 participants

participants (2)

Adrian Schröter
Stefan Botter