!SxCwoMDDLFxMJcVaKl:matrix.org

syslog-ng

182 Members
3 Servers

Load older messages


SenderMessageTime
26 Jul 2023
@_discord_565564561357275176:t2bot.ioczp Norm1 and norm2 are without parallelize(), para1 and para2 are with parallelize() 10:39:15
@_discord_565564561357275176:t2bot.ioczp For the single client scenario there is roughtly a 40% performance increase. 10:41:20
@_discord_565564561357275176:t2bot.ioczp * For the single client scenario there is roughly a 40% performance increase. 10:41:32
@_discord_565564561357275176:t2bot.ioczp When there are many clients, then there is a minimal performance decrease, most visible around 50 clients. 10:43:24
@_discord_565564561357275176:t2bot.ioczpsng43norm_vs_parallelize_vs_old.png
Download sng43norm_vs_parallelize_vs_old.png
10:46:24
@_discord_565564561357275176:t2bot.ioczp And here comes the "everything was better in the good old times" graph: same hardware, but older openSUSE and syslog-ng versions 🙂 10:46:24
@_discord_565564561357275176:t2bot.ioczp The parallelize tests were done on openSUSE Leap 15.5 and syslog-ng 4.3.0. 10:47:28
@_discord_565564561357275176:t2bot.ioczp The old measurement was done on the same AMD 5800X box, but openSUSE Leap 15.2 and syslog-ng 3.26. 10:48:11
@_discord_565564561357275176:t2bot.ioczp And a quick test with syslog-ng 4.0 (the oldest syslog-ng version I could find for openSUSE 15.5): 11:17:44
@_discord_565564561357275176:t2bot.ioczpimage.png
Download image.png
11:17:47
@_discord_565564561357275176:t2bot.ioczp Using the distro (complex) config and many connections, it's much faster than syslog-ng 4.3 on the same OS and hardware. 11:19:33
@_discord_1133825220814327928:t2bot.iokefiras_32705 joined the room.18:18:59
27 Jul 2023
@_discord_258949722205716480:t2bot.iogrg_2145 Hi,

I opened the issue https://github.com/syslog-ng/syslog-ng/issues/4566 about using the same python parser multiple times, causes syslog-ng to segfault on shutdown, I don't know what is the difference between mine and the one which Bazsi gave in https://github.com/syslog-ng/syslog-ng/pull/4552 but the latter does not segfaults.

I encountered it whem I was trying to add testcases for the original pr, into testdb so that we will have this tested, unfortenately the tcs are failing because of this. (not pushed yet)
07:50:17
@_discord_258949722205716480:t2bot.iogrg_2145 * Hi,

I opened the issue https://github.com/syslog-ng/syslog-ng/issues/4566 about using the same python parser multiple times, causes syslog-ng to segfault on shutdown, I don't know what is the difference between mine and the one which Bazsi gave in https://github.com/syslog-ng/syslog-ng/pull/4552 but the latter does not segfaults.

I encountered it whem I was trying to add testcases for the original pr, into testdb so that we will have this tested, unfortenately the tcs are failing because of this. (not pushed yet)
Edit I just checked it it happening on reload as well
07:55:53
@_discord_565564561357275176:t2bot.ioczp Having ivykis "officially" updated with your PR instead of patching it would be fantastic. 10:54:51
@_discord_565564561357275176:t2bot.ioczp I just ran into a problem on FreeBSD: if ivykis is installed on the system, compiling of syslog-ng fails, even if using --with-ivykis=internal 10:56:14
@_discord_565564561357275176:t2bot.ioczp Until now I compiled syslog-ng with internal ivykis, but for 4.3.0 I have to change ports to use ivykis from system, otherwise compilation might fail. 10:57:32
@_discord_395690247985561612:t2bot.iogokhanadev joined the room.13:47:04
28 Jul 2023
@_discord_839876246518366218:t2bot.iohofione Same on macOS, I created a bug for this and hopefully will be able to work on it next week 18:41:29
@_discord_914929598179975239:t2bot.iobazsi77 Can you post an error log? 19:14:06
@_discord_565564561357275176:t2bot.ioczp This is from FreeBSD, when Cy tried to compile syslog-ng with internal ivykis, while ivykis was also installed on the system: 19:16:15
@_discord_565564561357275176:t2bot.ioczp libtool: link: cc -Wshadow -Wno-stack-protector -Wno-unused-parameter -Wno-variadic-macros -O2 -g -Wno-initializer-overrides -O2 -pipe -pipe -fstack-protector-strong -fno-strict-aliasing -fstack-protector-strong -o modules/correlation/pdbtool/.libs/pdbtool modules/correlation/pdbtool/pdbtool-pdbtool.o -Wl,--export-dynamic ./lib/.libs/libsyslog-ng.so ./modules/correlation/.libs/libsyslog-ng-patterndb.a /export/wrkdir/amd64/export/freebsd/git/ports/sysutils/syslog-ng/work/syslog-ng-4.3.0/lib/.libs/libsyslog-ng.so -L/usr/local/lib /export/wrkdir/amd64/export/freebsd/git/ports/sysutils/syslog-ng/work/syslog-ng-4.3.0/lib/eventlog/src/.libs/libevtlog.so -L./lib/ivykis/src -livykis /export/wrkdir/amd64/export/freebsd/git/ports/sysutils/syslog-ng/work/syslog-ng-4.3.0/lib/secret-storage/.libs/libsecret-storage.so -L/usr/lib -lcrypto -lssl -lm -lrt -lgmodule-2.0 -lgthread-2.0 -lglib-2.0 -lintl ./lib/eventlog/src/.libs/libevtlog.so ./lib/secret-storage/.libs/libsecret-storage.so -lpcre2-8 -lregex -ldl -pthread -Wl,-rpath -Wl,/usr/local/lib
ld: error: undefined reference due to --no-allow-shlib-undefined: iv_work_pool_submit_continuation
referenced by ./lib/.libs/libsyslog-ng.so
cc: error: linker command failed with exit code 1 (use -v to see invocation)
gmake[4]: [Makefile:16089: modules/correlation/pdbtool/pdbtool] Error 1
gmake[3]:
[Makefile:27467: all-recursive] Error 1
gmake[2]: *** [Makefile:11770: all] Error 2
gmake[2]: Leaving directory '/export/wrkdir/amd64/export/freebsd/git/ports/sysutils/syslog-ng/work/syslog-ng-4.3.0'
19:16:18
@_discord_914929598179975239:t2bot.iobazsi77 It seems it is linking against the system ivykis 19:16:22
@_discord_914929598179975239:t2bot.iobazsi77 While we detected the required functionality in the bundled one. 19:16:41
@_discord_914929598179975239:t2bot.iobazsi77 It seems that the -L option for /use/local/lib comes first 19:17:58
@_discord_914929598179975239:t2bot.iobazsi77 If you move -L/usr/local/lib to somewhat later in the command line, does that help? 19:19:08
@_discord_914929598179975239:t2bot.iobazsi77 The issue is that the -L option for /usr/local/lib might be coming from another dependency, I am not sure what adds it 19:19:41
@_discord_914929598179975239:t2bot.iobazsi77 I know. I was just saying it might be difficult to get that -L to be later, when generated by configure. But I will check. 19:24:51
29 Jul 2023
@_discord_1009847941231554681:t2bot.ioharmonious_guava_50390 joined the room.08:29:44
1 Aug 2023
@_discord_1135757089633996853:t2bot.iotaotaowang_74689 joined the room.02:13:48

Show newer messages


Back to Room ListRoom Version: 10