4.1.4. hangs sometimes

[Clockwork]

Since 4.1.4 (?) I have the problem that the "litespeed" process hangs up sometimes (it shows 100% cpu @ top and doesn't do anything) without spawning new lsphp5 processes. I'm required to manually restart lsws to solve this problem.

There is also no error message in the error.log or stderr.log file.

[mistwang]

Please do a force reinstall of 4.1.4 to make sure getting the latest build. We fixed a bug in realtime statistics report that may cause 100% problem. Not sure it is the one you experiencing or not.
If it happens again, please try "strace -tt -T -p <pid_of_100_CPU_process>", for a little while before killing it. please send us the output for analysis.

[andreas]

Happened today with v4.1.6. Strace log:

20:38:49.555439 gettimeofday({1319395129, 555448}, NULL) = 0 <0.000007>
20:38:49.555467 epoll_wait(6, {{EPOLLIN, {u32=7, u64=7}}, {EPOLLIN, {u32=9, u64=9}}}, 10, 100) = 2 <0.000007>
20:38:49.555498 gettimeofday({1319395129, 555507}, NULL) = 0 <0.000007>
20:38:49.555526 epoll_wait(6, {{EPOLLIN, {u32=7, u64=7}}, {EPOLLIN, {u32=9, u64=9}}}, 10, 100) = 2 <0.000007>
20:38:49.555556 gettimeofday({1319395129, 555566}, NULL) = 0 <0.000007>
20:38:49.555584 epoll_wait(6, {{EPOLLIN, {u32=7, u64=7}}, {EPOLLIN, {u32=9, u64=9}}}, 10, 100) = 2 <0.000007>
20:38:49.555615 gettimeofday({1319395129, 555624}, NULL) = 0 <0.000007>
20:38:49.555643 epoll_wait(6, {{EPOLLIN, {u32=7, u64=7}}, {EPOLLIN, {u32=9, u64=9}}}, 10, 100) = 2 <0.000008>
20:38:49.555674 gettimeofday({1319395129, 555683}, NULL) = 0 <0.000007>
20:38:49.555702 epoll_wait(6, {{EPOLLIN, {u32=7, u64=7}}, {EPOLLIN, {u32=9, u64=9}}}, 10, 100) = 2 <0.000007>
20:38:49.555732 gettimeofday({1319395129, 555742}, NULL) = 0 <0.000007>
20:38:49.555760 epoll_wait(6, {{EPOLLIN, {u32=7, u64=7}}, {EPOLLIN, {u32=9, u64=9}}}, 10, 100) = 2 <0.000007>
20:38:49.555791 gettimeofday({1319395129, 555800}, NULL) = 0 <0.000007>
20:38:49.555819 epoll_wait(6, {{EPOLLIN, {u32=7, u64=7}}, {EPOLLIN, {u32=9, u64=9}}}, 10, 100) = 2 <0.000007>
20:38:49.555849 gettimeofday({1319395129, 555859}, NULL) = 0 <0.000007>
20:38:49.555877 epoll_wait(6, {{EPOLLIN, {u32=7, u64=7}}, {EPOLLIN, {u32=9, u64=9}}}, 10, 100) = 2 <0.000007>
20:38:49.555908 gettimeofday({1319395129, 555917}, NULL) = 0 <0.000007>
20:38:49.555936 epoll_wait(6, {{EPOLLIN, {u32=7, u64=7}}, {EPOLLIN, {u32=9, u64=9}}}, 10, 100) = 2 <0.000008>
20:38:49.555966 gettimeofday({1319395129, 555976}, NULL) = 0 <0.000007>
20:38:49.555995 epoll_wait(6, {{EPOLLIN, {u32=7, u64=7}}, {EPOLLIN, {u32=9, u64=9}}}, 10, 100) = 2 <0.000007>
20:38:49.556025 gettimeofday({1319395129, 556034}, NULL) = 0 <0.000007>
20:38:49.556053 epoll_wait(6, {{EPOLLIN, {u32=7, u64=7}}, {EPOLLIN, {u32=9, u64=9}}}, 10, 100) = 2 <0.000007>
20:38:49.556083 gettimeofday({1319395129, 556093}, NULL) = 0 <0.000007>
20:38:49.556112 epoll_wait(6, {{EPOLLIN, {u32=7, u64=7}}, {EPOLLIN, {u32=9, u64=9}}}, 10, 100) = 2 <0.000007>
20:38:49.556142 gettimeofday({1319395129, 556151}, NULL) = 0 <0.000007>
20:38:49.556170 epoll_wait(6, {{EPOLLIN, {u32=7, u64=7}}, {EPOLLIN, {u32=9, u64=9}}}, 10, 100) = 2 <0.000008>
20:38:49.556201 gettimeofday({1319395129, 556210}, NULL) = 0 <0.000007>
20:38:49.556229 epoll_wait(6, {{EPOLLIN, {u32=7, u64=7}}, {EPOLLIN, {u32=9, u64=9}}}, 10, 100) = 2 <0.000007>
20:38:49.556266 gettimeofday({1319395129, 556275}, NULL) = 0 <0.000007>
20:38:49.556294 epoll_wait(6, {{EPOLLIN, {u32=7, u64=7}}, {EPOLLIN, {u32=9, u64=9}}}, 10, 100) = 2 <0.000007>
20:38:49.556325 gettimeofday({1319395129, 556334}, NULL) = 0 <0.000007>
20:38:49.556353 epoll_wait(6, {{EPOLLIN, {u32=7, u64=7}}, {EPOLLIN, {u32=9, u64=9}}}, 10, 100) = 2 <0.000008>
20:38:49.556383 gettimeofday({1319395129, 556393}, NULL) = 0 <0.000007>
20:38:49.556412 epoll_wait(6, {{EPOLLIN, {u32=7, u64=7}}, {EPOLLIN, {u32=9, u64=9}}}, 10, 100) = 2 <0.000007>
20:38:49.556442 gettimeofday({1319395129, 556452}, NULL) = 0 <0.000007>
20:38:49.556471 epoll_wait(6, {{EPOLLIN, {u32=7, u64=7}}, {EPOLLIN, {u32=9, u64=9}}}, 10, 100) = 2 <0.000007>

Back to 4.0.20, once again...

[mistwang]

This issue is likely fixed in our 4.1.7 release. We got positive feedbacks from a customer experiencing similar problem.

[IrPr]

Unfortunately I got the same issue on 4.1.7 also!
None of 4.1.x series are stable on my busy servers and I'm forced to use 4.0.20 instead which is very more stable in compare of 4.1.x

I really don't get which change leads to this instability but I've to say that 4.1.x is not stable, at least on my busy servers
Seems that LS folks couldn't solve the issue after this long time

Anyway, It's preferred to use 4.0.20 which is really awesome without any similar issue! the only trade-off is AIO which is not implemented in 4.0

I'll looking forward to see if you can provide real fix

Regards

[Tony]

Yep we still see this in 4.1.7 at random as well. Just randomly happens on a server and restart lsws and issue is fixed then may not see it again only to have it pop up on another server.

[mistwang]

Thanks to a reliable bug report, we fixed a 100&#37; cpu bug which happens after the connection limit has been reached once.
It can be easily reproduced by set "Max SSL connections" to "0", then access a https web site, then access a normal web site.

Fix is in up coming 4.1.8 release.