4.1.12 broken

[raphidae]

I upgraded a server from 4.1.10 to 4.1.12 today and it completely broke php and the web console.

All php scripts result in 503, as does accessing the web console.

there are numerous unhelpful errors in stderr.log:

2012-04-05 23:18:03.006 [STDERR] lscgid: recv_fd() failed: No such file or directory
2012-04-05 23:18:03.007 [STDERR] lscgid: recv_fd() failed: No such file or directory
2012-04-05 23:18:14.947 [STDERR] lscgid: recv_fd() failed: No such file or directory

and some more in my admin error log:

2012-04-05 23:15:25.057 [INFO] [xx.44.172.93:54313-0#_AdminVHost] ExtConn timed out while processing.
2012-04-05 23:15:25.057 [INFO] [xx.44.172.93:54313-0#_AdminVHost] connection to [/tmp/lshttpd/admin_php.sock] on request #0, confirmed, 0, associated process: 1952543827, running: 0, error: Operation timed out!
2012-04-05 23:15:25.057 [NOTICE] [xx.44.172.93:54313-0#_AdminVHost] Suspicious request timeout, associated process: 1952543827, running: 0.

Reverting to 4.1.10 solved all problems instantly.

Additionally, since upgrading I've noticed a new file: /tmp/secret/lscgid.sock

I can't find any references to this file in any of the config files, so I'm unsure of the purpose. Other servers that still run 4.1.10 don't have this file.

I tried deleting it, but that results in:

stderr.log:2012-04-05 23:40:58.785 [STDERR] lscgid: failed to open secret file: /tmp/secret/lscgid.sock.724!
stderr.log:lscgid: secret is not set properly!

Reverting back to 4.1.10 did not remove the need for this file for some reason.

[raphidae]

Maybe I should have noted that this is on FreeBSD 8-STABLE. Are your testing procedures for FreeBSD the same as for Linux?

It seems every time I upgrade there is one issue or another. If you're not testing the FreeBSD builds the same as the Linux builds I'd like to know, since I'm paying the same as a Linux user.

[raphidae]

FYI: I just tried 4.1.11 to see whether that version works, and it doesn't have these issues. It seems to be new in 4.1.12.

[webizen]

stay with 4.1.11 for now.

[raphidae]

Originally Posted by webizen

stay with 4.1.11 for now.

Can you perhaps explain the /tmp/secret/lscgid.sock file in the meanwhile?

[webizen]

not seen /tmp/secret/lscgid.sock file in our lab following your upgrade path (4.1.10 -> 4.1.12 -> 4.1.11). try full stop/start and see if that clears.

[priestjim]

I get extremely high loads on 4.1.12. The error log says:

2012-04-10 13:15:44.416 [WARN] Remove looping fd: 50, event: 1
2012-04-10 13:15:44.465 [WARN] Remove looping fd: 797, event: 1
2012-04-10 13:15:44.521 [WARN] Remove looping fd: 822, event: 1
2012-04-10 13:15:44.730 [WARN] Remove looping fd: 747, event: 1
2012-04-10 13:15:44.863 [WARN] Remove looping fd: 833, event: 1
2012-04-10 13:15:44.913 [WARN] Remove looping fd: 840, event: 1
2012-04-10 13:15:45.079 [WARN] Remove looping fd: 139, event: 1
2012-04-10 13:15:45.128 [WARN] Remove looping fd: 494, event: 1
2012-04-10 13:15:45.180 [WARN] Remove looping fd: 522, event: 1
2012-04-10 13:15:45.241 [WARN] Remove looping fd: 723, event: 1
2012-04-10 13:15:45.271 [WARN] Remove looping fd: 509, event: 1
2012-04-10 13:15:45.409 [WARN] Remove looping fd: 529, event: 1
2012-04-10 13:15:46.577 [WARN] Remove looping fd: 851, event: 1
2012-04-10 13:15:47.018 [WARN] Remove looping fd: 73, event: 1
2012-04-10 13:15:47.425 [WARN] Remove looping fd: 859, event: 1
2012-04-10 13:15:47.743 [WARN] Remove looping fd: 887, event: 1
2012-04-10 13:15:49.231 [WARN] Remove looping fd: 319, event: 1

And strace-ing the litespeed process gives out A LOT of:

mincore(0x7ff7b9ac5000, 443819, 0x7fffc0296aa0) = -1 ENOMEM (Cannot allocate memory)

[webizen]

switch to 4.1.11 see if any difference.

[priestjim]

Did that in order to get my server to an operable state. With 4.1.11 I get no load issues.

[webizen]

stay with 4.1.11 for now until 4.1.12 becomes available in auto-update.