|
Author
|
Message
|
|
Veteran Member
Join Date: Jun 2006
Location: x-base.info
|

07-03-2006
, 16:57
Re: socket_recv() hangs
|
#1
|
yesterday i did everything i could to get a good test case. remember, the server already crashed before. it had the plugin running with a server check frequency of 60 seconds. it was checking 7 servers, 4 of them Dominion's servers and 3 are IP's that don't run a server, so they are down. it didn't crash a whole night.
one day later i was frustrated and restarted the server. 4 hours later or so it crashed.
now yesterday i set the check frequency to 5 seconds. so now 7 servers are checked every 5 seconds. guess what: the server ran stable the whole night.
so as long as you didn't test for at least 48 hours continously you can't say it works. this is exactly what makes this a debugging hell.
between times i even believed the crash problem had magically fixed itself or me and the other people who had crashes just all had hallucinations.
but then after almost 48 hours, the next crash. and then sometimes 30 minutes after server restarting a crash, and so on...
there is just ZERO system behind this.
btw i don't know whether i receive a blank packet somewhere but that really doesn't matter. there was a log_amx() directly after socket_recv() that was never called. so i can code what i want after this line, it won't be of any help.
anyway, thank you for investing time in trying to find the problem. i really start to run out of ideas, the only thing that remains to try is whether a different timeout value would help as suggested by Hawk. again, the problem with this is that i could test it now and after 3 days without crash think it works and then at day 4 get the server hung up again. i wish i would know something to encourage a crash to be able to make faster tests.
__________________
Got more than one HL1 (CS, DoD, NS, TS, TFC, HLDM...) server? Check:

|
|
|
|