!YjuQUNkcLAURsubBFF:matrix.org

Kubernetes

1171 Members
Container orchestration from Google http://kubernetes.io/136 Servers

Load older messages


SenderMessageTime
21 Oct 2021
@aceart.de:matrix.orgaceArt.de5min for an empty file :/ So thats not a common mistake? Okay, will try to find something out11:25:52
@rio:rio.systemscxz38Don't forget DNS! It's always DNS ;)11:34:04
@aceart.de:matrix.orgaceArt.deI am not using DNS at the moment. Just the external IP of the loadbalancer11:35:37
@aceart.de:matrix.orgaceArt.deit seems to be my storage mount making problems11:35:51
@neo:shdw.frHaley
In reply to @rio:rio.systems
Don't forget DNS! It's always DNS ;)
No. It's always the Network. and DNS is part of the network! :D
11:36:37
@aceart.de:matrix.orgaceArt.demaybe you are right. I can't even ping the IP11:44:44
@rio:rio.systemscxz38
In reply to @neo:shdw.fr
No. It's always the Network. and DNS is part of the network! :D
I know some people who would very much disagree 😉
11:47:30
@oki.arief:matrix.orgOki Arief cxz38: 11:47:46
@rio:rio.systemscxz38
In reply to @aceart.de:matrix.org
maybe you are right. I can't even ping the IP
Also, not all IPs will play nicely when you ping them.
11:47:55
@aceart.de:matrix.orgaceArt.de
In reply to @rio:rio.systems
Also, not all IPs will play nicely when you ping them.
I rather think my IP got blacklisted or something like that.
Neither ping nor opening in browser works
11:48:29
@neo:shdw.frHaley Does not matter... I worked on both sides, and usually the network stops dns from being reachable. Because, you barely touch the dns server once it is setup! 11:50:04
@ardimaul:matrix.orgOrraa
In reply to @aceart.de:matrix.org
opening an empty file (or any file at all) takes minutes through a simple loadbalancer. What can be the problem?
Ur internet must be slow to start handshake session, default connection almost all http server has 60s timeout
11:52:40
@neo:shdw.frHaley
In reply to @rio:rio.systems
I know some people who would very much disagree 😉
* Does not matter... I worked on both sides, and usually the network stops dns from being reachable. Because, you barely touch the dns server once it is setup!
11:53:34
@aceart.de:matrix.orgaceArt.de380mbit download. 60 up. Its not my internet. It is something on Azure/my cluster11:54:22
@ardimaul:matrix.orgOrraa changed their display name from Ardi to Orraa.11:54:24
@ardimaul:matrix.orgOrraaIf u can open it in 5 minutes, it's not about blacklisted, it must about bad routing or that I mention it, slow internet11:55:32
@neo:shdw.frHaley5 minute is too long for a timeout IMHO.11:58:02
@neo:shdw.frHaley60second is what it usually takes.11:58:14
@ardimaul:matrix.orgOrraaWhat if they got routing/open connection at 4.30 seconds.. Or they not defined http timeout, using direct library maybe for open http server12:01:03
@lucasmerlin:matrix.orglucasmerlin joined the room.19:01:26
@ddeth84:matrix.orgrequiem set a profile picture.21:31:54
@ddeth84:matrix.orgrequiem changed their display name from D to requiem.21:32:29
22 Oct 2021
@johnnyj32:matrix.org@johnnyj32:matrix.org joined the room.04:21:29
@johnnyj32:matrix.org@johnnyj32:matrix.org left the room.04:23:04
@forgotten_panda:matrix.orgforgotten_panda joined the room.04:27:16
@rpotter6298:matrix.orgRyan Pottermaybe a silly question, but can a container use CPU resources from multiple hardwares? Like, if I have 10 cores in each of 4 servers, is there any way to have a process which can effectively assign 40 threads to each of these cores?10:56:47
@ardimaul:matrix.orgOrraa
In reply to @rpotter6298:matrix.org
maybe a silly question, but can a container use CPU resources from multiple hardwares? Like, if I have 10 cores in each of 4 servers, is there any way to have a process which can effectively assign 40 threads to each of these cores?
Good questions, No
11:02:05
@ardimaul:matrix.orgOrraaBecause every node communicate with networking layer 2/3/4, cpu can't be top of that, different level speed Unless u figure new hardware and software, and direct connect that cpu between nodes11:04:14
@rpotter6298:matrix.orgRyan Potter
In reply to @ardimaul:matrix.org
Because every node communicate with networking layer 2/3/4, cpu can't be top of that, different level speed
Unless u figure new hardware and software, and direct connect that cpu between nodes
okay, so I am guessing instead to accomplish a similar concept, you would have pods that run on each node which receive "orders" from a manager that is splitting up the tasks? I'm knowingly way out of my depth here, kinda trying to figure out what I need to learn more about
11:09:08
@bobbyd0g:matrix.orgbobbyd0g set a profile picture.15:25:09

There are no newer messages yet.


Back to Room List