Closed negz closed 6 years ago
Hi @negz. Thanks for your PR.
I'm waiting for a kubernetes-incubator or kubernetes member to verify that this patch is reasonable to test. If it is, they should reply with /ok-to-test
on its own line. Until that is done, I will not automatically test new commits in this PR, but the usual testing commands by org members will still work. Regular contributors should join the org to skip this step.
I understand the commands that are listed here.
/assign mtaufen
Ping.
I've been running this in production for a little over a week. No increase in issues, though ironically ip-masq-agent now crashes and restarts because the iptables lock is taken rather than due to an obfuscated iptable-restore
issue. :)
Hi @negz, sorry for the delayed response, I haven't owned this component for several years so I don't really pay close attention to it.
Rohit Ramkumar (@rramkumar1), Jing Ai (@jingax10) are the current owners, per the readme. Can one of you two help review this?
/ok-to-test
/assign
I've tried to keep the commits relatively clean, but I was somewhat foiled by godep and my unfamiliarity with it.
I've stepped into various issues with Godep too. We should probably switch to dep at some point to make our life easier.
I've been running this in production for a little over a week. No increase in issues, though ironically ip-masq-agent now crashes and restarts because the iptables lock is taken rather than due to an obfuscated iptable-restore issue. :)
Great that you have been consistently testing this. Given how ip-masq-agent operate, I'm guessing restarts and crashes is not ideal but should be fine?
Given how ip-masq-agent operate, I'm guessing restarts and crashes is not ideal but should be fine?
That's a good question! I asked about that in #23 and am curious what your thoughts are. At Planet Labs we run ip-masq-agent
in kube-system
, and we send a low priority, business hours only, oncall alert when any container in kube-system
restarts 'too much'.
So, in short, there's no actual impact from ip-masq-agent
restarts for us apart from alert noise. I raised #23 to determine whether this restart was by design. If it is, we can simply not alert when ip-masq-agent
restarts. If it isn't by design it makes sense to me to change ip-masq-agent
's behaviour to log or similar rather than crashing when it encounters an issue.
@negz Thanks for raising that, I agreed with what you proposed. Sent https://github.com/kubernetes-incubator/ip-masq-agent/pull/24 for this.
Testing if I can approve this after https://github.com/kubernetes-incubator/ip-masq-agent/pull/25. /approve
[APPROVALNOTIFIER] This PR is APPROVED
This pull-request has been approved by: MrHohn, negz
The full list of commands accepted by this bot can be found here.
The pull request process is described here
...which required updating various other things. I've tried to keep the commits relatively clean, but I was somewhat foiled by
godep
and my unfamiliarity with it.The goal here is to (potentially) fix #21 by using a version of
pkg/util/iptables
that can collaborate with other frequent callers of iptables, such as kube-proxy, per https://github.com/kubernetes/kubernetes/pull/44895.