hi, Rainer Gerhards wrote: > Thanks, I now received the log files. And, of course, they made > immediately clear what is going on ;) What you experience is flow > control kicking in. Let me first elaborate a bit: > > Rsyslog knows inputs which can be delayed a bit (like tcp receiver AND > the local socket) and those who can not (like UDP). The idea is that we > can slow down senders where this does not result in loss of messages, > leaving room inside the queue on busy systems for those where we can not > do that. We had cases where not delaying caused a lot of grief. > > There are also two watermarks, the low and high delay mark. If the low > delay mark is reached, a delayable source is delayed for one second. It > is at 70% of the (in-memory) queue capacity. Again, this is all done for > some very good reasons. > > However, when I introduced these features, I did not make them > configurable. Also, I did classify the local unix socket as one who can > be delayed. In your scenario, this does not stop the system, but once > the local log socket is filled up, it needs to wait one second for each > message before it is pulled. > > Again, the rate limiting logic is there for good reasons. So what I now > see is that I need to add a few config statements to allow to control it > according to your needs. I am also unsure if I should re-classify > imuxsock as an input which cannot be delayed - but that also can have > some drawbacks. Maybe the best thing to do is make its state > configurable and use "not to be delayed" as default.
ok, after reading through your explanation, i (almost) understand the delay mechanism and that at present it is not configurable. almost, because i do not get the part where you talk about the "70% of the (in-memory) queue capacity". anyways, how does this relate to the original problem, that the following code stalls the logging process. >>> #! /bin/sh >>> for ((i=1;i<=10000;i+=1)); do >>> logger Log line $i >>> done is this a capacity problem? is this another problem? i really consider this a critical bug, as i yesterday ran out of diskspace on my remote-rsyslog-and-mysql-on-one-host-testbed and the rsyslog-client, which logs locally and to the remote destination, stalled completely. cheers, raoul ps: i am on 3.16.1-1~bpo40+1 -- ____________________________________________________________________ DI (FH) Raoul Bhatia M.Sc. email. [EMAIL PROTECTED] Technischer Leiter IPAX - Aloy Bhatia Hava OEG web. http://www.ipax.at Barawitzkagasse 10/2/2/11 email. [EMAIL PROTECTED] 1190 Wien tel. +43 1 3670030 FN 277995t HG Wien fax. +43 1 3670030 15 ____________________________________________________________________ _______________________________________________ rsyslog mailing list http://lists.adiscon.net/mailman/listinfo/rsyslog
