View Issue Details
| ID | Project | Category | View Status | Date Submitted | Last Update |
|---|---|---|---|---|---|
| 0000721 | unreal | ircd | public | 2003-02-12 23:50 | 2003-04-11 17:14 |
| Reporter | Tiburon | Assigned To | syzop | ||
| Priority | normal | Severity | major | Reproducibility | always |
| Status | closed | Resolution | unable to duplicate | ||
| Summary | 0000721: Select Error Causes High CPU Usage [waiting] | ||||
| Description | [5:37pm] -Irc.SentinelScript.Net- select Irc.SentinelScript.Net[64.246.0.230]:Bad file descriptor Errors like that happen repeatedly and cause the ircd to pause for a few seconds and are making the ircd CPU usage to average around 30 percent. | ||||
| Steps To Reproduce | Run the ircd, get a decent amount of users. | ||||
| Additional Information | Also beta14 randomly crashes on some of the other servers on rehash | ||||
| 3rd party modules | |||||
|
|
also, the scan modules as well as ident have been disabled (we use opsb) |
|
|
Hi, #0000708 has the same problem. It will indeed cause a sleep() for 10 seconds. Looks like it's Linux 2.4.x related (altough probably an error on our side). About the crash on rehash, if it really crashed (so you have a core file) report it in a different bug along with a stack trace: -- gdb src/ircd core.file bt -- I might take a look at the select() problem but it's not very easy to trace (it happens because we somewhere didnt check for an error and therefore it will complain when select() is called with the bad filedescriptor). |
|
|
Btw, you don't get another error right before this one? |
|
|
I don't know how often you get this, but can you do (from your unrealircd dir): strace -o blah.log -p `cat ircd.pid` and ctrl+C as soon as you get an error notice again and mail it to me: tail -n 20000 blah.log|mail -s "strace log" [email protected] (or mail the full blah.log if you were slow in CTRL+c'ing, but then mail it [g]zipped or something ;P). Thanks! |
|
|
I've looked at the source but was unable to trace the problem, so I really need some strace output :(. |
|
|
Help! ;P *reminder* |
|
|
FYI this cannot be fixed in b15 (~1 or 2 weeks) without help from strace logs. |
|
|
it doesnt seem to happen at all on my bsd servers, scratches his head |
|
|
I cannot help at all without those strace output. And even then it might still be hard to trace. Tracing a bug caused by something which happend earlier at an unknown location in the code isn't exactly easy (another example of such a thing is memory corruption). Hopefully someone (else) will give me this output. |
|
|
Just for the record: this bug will not be fixed in beta15... Maybe in beta16, who knows :P. |
|
|
need data.. |
