Bug#1036213: apache2: frequent SIGSEGV in mod_http2.so (purge_consumed_buckets)
Le 19/05/2023 à 18:11, Stefan Eissing a écrit : Hi Bastien, Hello I am looking into the failures of mod_proxy_http2 and seeing an issue with connection reuse and subsequent handling of requests. Going back to the original issue, do you still see those with mod_h2 v2.0.15? That would be interesting to know. No, I had only 2 segfaults since I upgraded to 2.0.15, and none is in purge_consumed_buckets : - proxy_pass_brigade - ap_save_brigade I do not know if just relying on mod_proxy_http in your setup would be possible. But it could help us isolating issues. I don't think so ... The knot-resolver I forward DOH requests to seems to only support HTTP/2 ; if I change it to https://, dog fails to resolve on it : Error [http]: Nameserver returned HTTP 502 (Proxy Error) Regards, -- Bastien Durel
Bug#1036213: apache2: frequent SIGSEGV in mod_http2.so (purge_consumed_buckets)
Hi Bastien, I am looking into the failures of mod_proxy_http2 and seeing an issue with connection reuse and subsequent handling of requests. Going back to the original issue, do you still see those with mod_h2 v2.0.15? That would be interesting to know. I do not know if just relying on mod_proxy_http in your setup would be possible. But it could help us isolating issues. Kind Regards, Stefan > Am 19.05.2023 um 15:04 schrieb Bastien Durel : > > Le jeudi 18 mai 2023 à 15:04 +0200, Bastien Durel a écrit : >> Le 18/05/2023 à 14:41, Stefan Eissing a écrit : >>> Did you have an warning message like "AH03516: unexpected NN >>> streams in hold" at that time in out error log? >> >> No (grepping AH03516 in *.log returns nothing (nor does "streams in >> hold")) >> > Hello. I got another one in ap_save_brigade() > > -- > Bastien >
Bug#1036213: apache2: frequent SIGSEGV in mod_http2.so (purge_consumed_buckets)
Le jeudi 18 mai 2023 à 15:04 +0200, Bastien Durel a écrit : > Le 18/05/2023 à 14:41, Stefan Eissing a écrit : > > Did you have an warning message like "AH03516: unexpected NN > > streams in hold" at that time in out error log? > > No (grepping AH03516 in *.log returns nothing (nor does "streams in > hold")) > Hello. I got another one in ap_save_brigade() -- Bastien output.log.bz2 Description: application/bzip