[06:19:05] <_joe_>	 I have noticed nowadays puppet runs take up to 2 minutes
[06:29:08] <_joe_>	 on like the deployment server
[06:29:15] <_joe_>	 any idea what's causing it to be so slow?
[06:39:43] <jayme>	 yes
[06:39:46] * jayme hides
[06:40:23] <jayme>	 I can imagine it's the amount of pki certificates that get refreshed every now and then
[07:00:15] <_joe_>	 jayme: I doubt it, but yeah if we use puppet to drive cert renewal then it could be one of the reasons
[07:00:23] <_joe_>	 but puppet was also slow on an appserver
[07:02:08] <jayme>	 that might be something different then...on deployment hosts it really is quite a bunch of certs for k8s users which means a lot of sequential calls to pki if they need refresh
[12:30:24] <claime>	 slyngs: Heads up, I'm about to send 1% of traffic to mw-on-k8s (up from .5% last week)
[12:45:39] <slyngs>	 claime: Exciting :-)
[12:45:53] <slyngs>	 I'm back btw
[12:46:17] <claime>	 slyngs: A whopping 85rps total!
[12:46:36] <slyngs>	 How will Kubernetes cope ;-)
[12:47:09] <TheresNoTime>	 (very exciting!)
[12:49:37] <_joe_>	 the total is well above 100 rps
[12:49:43] <_joe_>	 https://grafana.wikimedia.org/d/Og_UOWpGz/application-layer-rps?orgId=1&viewPanel=20
[12:50:53] <claime>	 _joe_: -int doesn't count :D
[12:51:09] <claime>	 And yes, I was only looking at eqiad, not codfw
[12:51:20] <_joe_>	 claime: well it counts when talking about our overall RPS numbers :)
[12:52:32] <claime>	 _joe_: What I mean is since we haven't migrated anything else to it for a while, its usage is holding steady, but I see your point
[12:53:05] <apergos>	 exciting regardless!
[14:34:43] <cwhite>	 moritzm: puppet is disabled on logstash1025 for a while now.  Do you know when it can be re-enabled?
[14:43:09] <moritzm>	 sorry, I thought I had already done, that. I'll re-enable shortly
[15:05:44] <vgutierrez>	 hmmm I'm having some issues with clang++ on bookworm
[15:05:50] <vgutierrez>	 (and cmake)
[15:06:11] <vgutierrez>	 basically this check check_symbol_exists(clock_gettime time.h FOLLY_HAVE_CLOCK_GETTIME) is failing
[15:06:50] <vgutierrez>	 and yeah.. libc6-dev is there
[15:09:50] <vgutierrez>	 and it doesn't happen with g++ 12
[15:18:31] <moritzm>	 than it sounds like an issue with libc++, the STL implementation used by clang?
[15:26:48] <vgutierrez>	 hmmm fixed
[15:27:13] <vgutierrez>	 for some stupid reason it was trying to use GCC to build C code (including those checks) and clang to build C++ code
[15:27:18] <vgutierrez>	 (folly code itself is C++)
[15:27:56] <vgutierrez>	 moritzm: thanks for pointing me into the right direction :)
[15:32:34] <moritzm>	 yw :-)
[15:44:13] <_joe_>	 vgutierrez: oh my, are you building folly, thrift and all that bullshit? Ifeel for you
[15:44:22] <_joe_>	 been there, done that multiple times :)
[15:44:41] <vgutierrez>	 trying to get katran up & running with just one compiler
[15:45:03] <_joe_>	 vgutierrez: yeah...
[16:59:32] <akosiaris>	 haven't seen mentions of folly for a while
[16:59:43] <akosiaris>	 it's making a come back via katran ...
[19:32:28] <inflatador>	 shouldn't it be 'molly' now? :P