Secure Shell consuming 4G+ of memory, triggering OOMs |
|||||
Issue descriptionhttps://listnr.corp.google.com/report/72847119362 The system is memory constrained, and Chrome can't find any tabs to kill to make up the gap. In the list of tabs/extensions/apps from the low memory handler, Secure Shell really sticks out: Extension [Secure Shell|Secure Shell|Secure Shell] 4332 MB private, 4337 MB shared, 2682 MB swapped, 69 FDs open of 2048 Manu, Is this very reproducible for you? It seems crazy to me that the memory usage by that extension is so high.
,
Sep 11 2017
,
Sep 15 2017
Oof, that's bad. I'd be interested in where all that memory is going. Since SecureShell is (mostly) a web app with some extension-y bits, I wonder if this is extensions related or if there's just a JS leak in SecureShell somewhere.
,
Sep 15 2017
in order to leak 4GB of JS data, you'd have to be catting a lot of content into the scrollback, or to hit some kind of infinite loop bug what version of Secure Shell are you using there ? can you make sure you've upgraded to at least 0.8.36.12 ?
,
Sep 15 2017
From the feedback report: pnhechapfaindjhompbnflcldabbghjo : Secure Shell : version 0_8_36_10
,
Sep 15 2017
yeah, def upgrade to .12. i fixed one edge case infinite loop, but that would have manifested itself as a hang, so you should have noticed by the app no longer responding :). and not sure that loop would have caused memory inflation.
,
Sep 15 2017
So you're basically saying that it would have been good to upgrade to a version that has a change that most likely does not match the reported buggy behavior? ;-P
,
Sep 15 2017
I assume extensions update themselves without manual intervention? I'll keep an eye out for something similar again in the future. Thanks!
,
Sep 15 2017
i don't know what you're trying to say. the latest release has fixes for known bugs. we don't know what bug you hit such that your memory usage exploded. but if you upgrade to the latest version and you no longer have a problem, then what's the point in triaging/debugging further ?
,
Sep 15 2017
Oh, it's not easily reproducible (see comment #1) and sounds bad enough to try and gather as much data as possible from that single occurrence? But then again I won't be the one doing that ;-P Just trying to provide useful data.
,
Sep 22 2017
assigning to vapier@ for further triage (not sure if you want to close this out or not).
,
Sep 22 2017
Sure, we can close, I just thought it might be useful. But it's probably not very actionable.
,
Sep 22 2017
if it's reproducible with .12+, then we can look into it.
,
Sep 22 2017
I'll keep an eye out for this. Thanks. |
|||||
►
Sign in to add a comment |
|||||
Comment 1 by manucornet@chromium.org
, Sep 11 2017