[opensuse-buildservice] obs 2.5.1 as local installation - a first glance
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
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
Hi Adrian, On Tue, 22 Apr 2014 11:14:21 +0200 "Adrian Schröter" <adrian@suse.de> 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
On Dienstag, 22. April 2014, 15:55:49 wrote Stefan Botter:
Hi Adrian,
On Tue, 22 Apr 2014 11:14:21 +0200 "Adrian Schröter" <adrian@suse.de> 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
Hi Adrian, On Tue, 22 Apr 2014 16:02:13 +0200 "Adrian Schröter" <adrian@suse.de> 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
participants (2)
-
Adrian Schröter -
Stefan Botter