Hey team, What are the trade-offs between running agent as init and running it under systemd. There is this interesting bug: https://bugzilla.redhat.com/show_bug.cgi?id=1890681 where the agent gets OOM-killed. That manifests itself with somewhat hard-to-diagnose symptoms. I opened https://github.com/kata-containers/kata-containers/issues/1111. The solution Qian suggested was to set oom_score_adjust. I investigate that. But that made me wonder about the trade-off between using systemd and agent-as-init. Do we have this documented somewhere? Thanks Christophe
On 2020/11/13 17:39, Christophe de Dinechin wrote:
Hey team,
What are the trade-offs between running agent as init and running it under systemd.
There is this interesting bug: https://bugzilla.redhat.com/show_bug.cgi?id=1890681 where the agent gets OOM-killed. That manifests itself with somewhat hard-to-diagnose symptoms. I opened https://github.com/kata-containers/kata-containers/issues/1111.
The solution Qian suggested was to set oom_score_adjust. I investigate that. But that made me wonder about the trade-off between using systemd and agent-as-init. Do we have this documented somewhere?
I don't think we have explicitly documented it anywhere. Would you mind starting one? The trade-off can be identified from several aspects including resource footprint, stability and functionality. Cheers, Tao -- Into something rich and strange.
What are the trade-offs between running agent as init and running it under systemd.
There is this interesting bug: https://bugzilla.redhat.com/show_bug.cgi?id=1890681 where the agent gets OOM-killed. That manifests itself with somewhat hard-to-diagnose symptoms. I opened >https://github.com/kata-containers/kata-containers/issues/1111.
The solution Qian suggested was to set oom_score_adjust. I investigate that. But that made me wonder about the trade-off between using systemd and agent-as-init. Do we have this documented somewhere?
I don't think we have explicitly documented it anywhere. Would you mind starting one? The trade-off can be identified from several aspects including resource footprint, stability and functionality.
We do have https://github.com/kata-containers/documentation/wiki/Kata-Images, but it looks like it needs a refresh given all the recent changes. Cheers, James --- https://katacontainers.io/ | https://github.com/kata-containers Open Source Technology Center Intel Corporation (UK) Ltd. - Co. Reg. #1134945 - Pipers Way, Swindon SN3 1RJ. --------------------------------------------------------------------- Intel Corporation (UK) Limited Registered No. 1134945 (England) Registered Office: Pipers Way, Swindon SN3 1RJ VAT No: 860 2173 47 This e-mail and any attachments may contain confidential material for the sole use of the intended recipient(s). Any review or distribution by others is strictly prohibited. If you are not the intended recipient, please contact the sender and delete all copies.
On 16 Nov 2020, at 10:11, Hunt, James O <james.o.hunt@intel.com> wrote:
What are the trade-offs between running agent as init and running it under systemd.
There is this interesting bug: https://bugzilla.redhat.com/show_bug.cgi?id=1890681 where the agent gets OOM-killed. That manifests itself with somewhat hard-to-diagnose symptoms. I opened >https://github.com/kata-containers/kata-containers/issues/1111.
The solution Qian suggested was to set oom_score_adjust. I investigate that. But that made me wonder about the trade-off between using systemd and agent-as-init. Do we have this documented somewhere?
I don't think we have explicitly documented it anywhere. Would you mind starting one? The trade-off can be identified from several aspects including resource footprint, stability and functionality.
We do have https://github.com/kata-containers/documentation/wiki/Kata-Images, but it looks like it needs a refresh given all the recent changes.
OK. I'll experiment with both configurations and update the documentation with the findings.
Cheers,
James --- https://katacontainers.io/ | https://github.com/kata-containers Open Source Technology Center Intel Corporation (UK) Ltd. - Co. Reg. #1134945 - Pipers Way, Swindon SN3 1RJ. --------------------------------------------------------------------- Intel Corporation (UK) Limited Registered No. 1134945 (England) Registered Office: Pipers Way, Swindon SN3 1RJ VAT No: 860 2173 47
This e-mail and any attachments may contain confidential material for the sole use of the intended recipient(s). Any review or distribution by others is strictly prohibited. If you are not the intended recipient, please contact the sender and delete all copies.
participants (3)
-
Christophe de Dinechin
-
Hunt, James O
-
Peng Tao