- 20 Sep, 2011 28 commits
-
-
Poul-Henning Kamp authored
test failures seems to be timeouts on stressed machines. Teport timeouts more clearly.
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
two separate timestamps just happen to be the same, just make them the same.
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
VSL record. Still not quite sure how they happen, but we shouldn't hang no matter what.
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
the same address again, even with hinting.
-
Poul-Henning Kamp authored
-
-
Poul-Henning Kamp authored
pretend nothing else happened.
-
Poul-Henning Kamp authored
-
Andreas Plesner Jacobsen authored
-
Poul-Henning Kamp authored
this should solve issues where the c2-v1 connection times out before it ever got used.
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
fd's early in the case of a timeout. Fixes #1023
-
-
Poul-Henning Kamp authored
-
Tollef Fog Heen authored
Apache (and nginx) uses 8k, so use the same default size to avoid people being surprised by more 413 responses than necessary. Fixes: #1016
-
Tollef Fog Heen authored
Thanks to cleberjsantos for initial patch which this is based on. Fixes: #976
-
Poul-Henning Kamp authored
are no lingering bits. Convert an assert to a failure.
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
-
- 19 Sep, 2011 6 commits
-
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
Fixes #1022
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
steps the clock... Fixes #1022
-
Poul-Henning Kamp authored
Make pipe mode use the acct_tmp like everybody else.
-
Poul-Henning Kamp authored
-
- 18 Sep, 2011 6 commits
-
-
Poul-Henning Kamp authored
Make one single thread, which creates thread pools, and since that is not particularly ardous work, also have it maintain the stats gauge of number of sessions queued. (Removing pools should now be possible, but the code is not there yet.) Each pool has its own herder thread, which creates and destroys threads as required, and updates stats for dropped and queued sessions. Sanitize VSC fields associated with this area.
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
Schedule sessions from busy-lists and from the waiter via SES_Schedule(), to Pool_Schedule() in the right pool.
-
Poul-Henning Kamp authored
Store the data from accept(2) on the worker->ws, we only need it for a brief moment until the worker configures a session with it. If we fail to allocate a session after accepting the connection, we silently close the connection again: Presumably it is a DoS situation. (false positive: Extremely popular busy objects) Make three clearly defined counters for sessions: Accepted, Dropped and failed (accept(2) failure). Lots of spit&polish.
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
Also check if the http_max_hdr param changes. Describe http_max_hdr as per pool. Add SES_DeletePool() function.
-