Sections
You are here: Home » Development » Testing » Test Results » Issue Tracker » [OpenTox website performance]

#6 — [OpenTox website performance]

State Tested and confirmed closed
Area Process
Issue type Feature
Severity Important
Submitted by Jeliazkov Vedrin
Submitted on Dec 19, 2009
Responsible Micha Rautenberg
Return to tracker
Last modified on Jul 14, 2013 by Micha Rautenberg
I've noticed huge performance degradation of the OpenTox website, which occurs every day of the week (except Saturday) between aprox. 01:00 GMT and 03:30 GMT. This might seriously hamper OpenTox website visibility and usability in other timezones, e.g. USA, etc... The response time is very long and frequently results in timeouts, which make any reasonable use of the website during this time frame close to impossible. Some further investigation has revealed that this problem is reflected also at ICMP level (simple ping):

http://ambit.uni-plovdiv.bg[…]arget=IPv4.Normal.OpenTox-N

In particular, please have a look at the "Last 72 hours" and "Last 7 days" graphs and observe the latency peaks which are recorded. Also please note that the application level latency during this time frame is a lot higher than the recorded ICMP round-trip time.

It looks like the server performs some scheduled (cron?) jobs which exhaust its resources. It might be wise to check:

(1) what are these jobs doing;
(2) whether they're really needed;
(3) should they run every day;

...and eventually check what are the options to upgrade the server and increase its performance if the answer to (3) is positive.
Steps to reproduce:
Try browsing http://www.opentox.org/ between 01:00 and 03:30 GMT on every day, except Saturday.
Added by Micha Rautenberg on Dec 19, 2009 01:49 PM
Issue state: UnconfirmedResolved
(1) there are backup jobs that runs every day 1:14 CET.
(2) yes
(3) sure

there are backup jobs during this time. Normally there should be a short period to create the backup files(1/2 - max 1 hour). After that there is a copy (synchronization) action to a backup-server, this action takes up to 2 hours. The backup server is a net-internal ftp server at Hetzner.
There is one full backup on Thursday and incremental backups on every other day the week.
I check and try to optimize or change the backup strategy before Xmas.
Added by Micha Rautenberg on Dec 19, 2009 01:50 PM
Issue state: ResolvedConfirmed
not resolved. it is confirmed.
Added by Micha Rautenberg on Dec 23, 2009 04:06 PM
Severity: CriticalImportant
changed options for the rsync transmission
give it a try during the next 2 weeks
Added by Jeliazkov Vedrin on Dec 31, 2009 02:37 PM
Issue state: ConfirmedResolved
The new settings seem to have resolved the issue, thanks!
Added by Jeliazkov Vedrin on Dec 31, 2009 02:38 PM
Issue state: ResolvedTested and confirmed closed

No responses can be added.