I wrote a shell-script to warn me, when SWAP is used over
a certain limit.
When I restarted mySQL and Apache everything went on well again.
It is hard to track down, which Oxid-template is causing this.
We have a mySQL-monitoring tool, which showed my at least the database,
causing this. The other stuff is logfile searching, searching, searching. ;-)
Not a nice job.
Uwe
| Hi again, currently 0.14, 0.25, 0.27 I can not say what it was when the machine crashed on saturday. We had the same situation a few days ago (4.5.2011 i think) when 50 nearly parallel querys came from a newly set facebook-comment that linked to shop 3. I have some statements of the technichian that analysed the incident: 1) [Die Datenbank sortiert etwas und wird nicht fertig, vmtl. für eine Ausgabe, dann der Neustart] 2) [Hier funktioniert PHP aufgrund von Speicherproblemen offenbar nicht mehr richtig, offenbar zuviele gleichzeitige Verbindungen] 3) [Zum Zeitpunkt des Restarts liefen 451 Instanzen von verschiedenen PHP-Skripten (welche, lässt sich leider nicht mehr ermitteln).] 4) [52 identische parallel gestellte Anfragen an den Server. Mglw. haben diese Anfragen die Datenbank lahmgelegt, was einige Zeit später zur Speicherknappheit führte] we had quite heavy load (munin-graphs) and a sudden peak of RAM-allocation wich led to RAM-shortage and finaly to Swaping a lot of pages (sudden increase of swap-calls and inode-consumption) After a few minutes and the restart of the box everything was fine again. On this saturday a similar situation: Box running at 80 to 98% load for quite a time already. and then munin shows the sudden rise of commited RAM (shooting up to 4.15 GB) and then the rise of Swap-consumption and inodes again. Result: no response of any shops. Not even HTML-Files were delivered and no response to SSH-commands. REBOOT, waiting a few minutes and everything was fine again - load returning to more than 90% for a while, but the machine remained stable until now. Best Regards / Mit freundlichem Gruß, Kai Gazmaga VektorDesign - Online-Shops und Programmierung
Am 17.05.2011 09:41, schrieb Uwe Degenhardt: |
to to forums-post you sent. Maybe it was misunderstandable.
But you have plenty of variants in at least two shops.
My guess is that those could cause the trouble before you upgraded.
What is your average load if you type: "top"
under linux ?
Uwe
--
SD Solutions
Lange Zeile 27
85435 Erding
Tel.: +498122903791
Fax.: +498122903792
eMail: [email protected]
Web: www.sdsolutions.de
I am really surprised, that your virtual box is still running. ;-) Especially if you go with variants. I have had a customer recently with a lot of variants who alone almost crashed our shared hosting box. Although I have to admit, it was a 4.2.x I currently do no have 15k visitors a day!!! I am running around 1300 to 1800 in 3 Main-Shops and a few additional visitors in 3 further shops. The XEN-Box had 1,5 GHz and 1.5 GB RAM (one core in the V-Box) on a quite up-to-date XEON-whatever-Host-Machine. I now upgraded to 4 cores, 1.5 GHz each and 4 GB RAM in the V-Box to bridge the time until the decision for the dedicated hardware is made and the server is fully configured and available. I wait eagerly for the munin-results of the next days to see what happens with the upgraded system. |
_______________________________________________ dev-general mailing list [email protected] http://dir.gmane.org/gmane.comp.php.oxid.general
