Sent from my BlackBerry® wireless device -----Original Message----- From: Peter Wake <[email protected]> Date: Tue, 26 Jun 2012 08:13:16 To: '[email protected]'<[email protected]> Reply-to: [email protected] Subject: RE: [Pound Mailing List] SSL performance
Hi Martin You might want to check out this thread: http://www.apsis.ch/pound/pound_list/archive/2012/2012-05/1337789033000#1337789033000 The key line in this is: # Increase the threads to 4096 - we tried 16000 but it didn't like it Threads 4096 I think it's due to pound's revised threading model, which started in version 2.5. Regards, Peter > -----Original Message----- > From: [email protected] [mailto:[email protected]] > Sent: 26 June 2012 08:33 > To: [email protected] > Subject: [Pound Mailing List] SSL performance > > Hello, > > I'm using pound 2.5 and am experiencing the following behavior when > doing a jmeter stresstest; > > Pound starts building up threads to around 900 (according to lsof) and > then suddenly drops to around 250. At this time jmeter starts showing > errors. After the drop the number increases again. Depending on the > jmeter parameters this behavior occurs several times. > > My setup; Pound is running SSL and behind it we have 2 varnish > instances. > When addressing Varnish directly everything works perfectly. > > We're using debian stable on a 8 core system with 24GB memory so I > don't expect any problems on that department. > > After a suggestion of Robert I upped the filedescriptors to 10240 so > that shouldn't be a problem either. > > Anyone having an idea what could be wrong or how I could start figuring > out what the problem might be ? > > -- > Tin will not rust or give in but melts directly when heated =B-) > > > -- > To unsubscribe send an email with subject unsubscribe to > [email protected]. > Please contact [email protected] for questions. -- To unsubscribe send an email with subject unsubscribe to [email protected]. Please contact [email protected] for questions. N‹§²æìr¸›zǧu©Þ™¨¥Â+a²æãyËnžË›±Êâmëh¦‹§uªlŠÇ!>Wš±ç(žÖœ¶º,z©²+ …ú+ªç¬¶*'
