Restarting ipfs daemon sometimes freezes the browser

[magik6k]

When restarting the local daemon (killing it and starting back after 10-30s), the browser will occasionally freeze for a few seconds.

ipfs-companion 2.1.0
Firefox 57.0.4
os/wm: Arch Linux/bspwm

[lidel]

I suspect this got fixed in past weeks.
Are you able to re-check with v2.2.0beta1 (from Development Channel)?
If the issue is still present, are you running multiprocess (about:support → "Multiprocess Windows")?

[magik6k]

So I've been running v2.2.0beta1 for some time on FF 58 now, and it is still an issue. Yes, it's running multiprocess. The issue is still there.

[lidel]

Thank you, I am looking into this as a part of stabilization efforts before v2.2.0 release. It is on our TODO list.
(FYI we should have a new beta within a week)

[lidel]

I am following up on things and I am unable to replicate this.

@magik6k any change with latest version? ('beta' and 'stable' have the same code right now)

[magik6k]

on 2.2.0.8720 - I've had some hangs but it feels like they happen less often now. I've also noticed that I get these hangs when using 'Copy Public Gateway URL' from the context menu (might not be related to this issue).

[lidel]

Yeah, copying should work instantly (it does not hit API), but.. are you having issues with pinning by any chance?
Also:
Is your node's HTTP API on the same machine or are you using a remote one?
Slow/old laptop?

[magik6k]

are you having issues with pinning by any chance?

The pin button in the extension doesn't seem to do anything for me. (pin doesn't show up in ipfs pin ls -t recursive, there is no feedback but the browser doesn't hang)

Is your node's HTTP API on the same machine or are you using a remote one?

It's on the local machine

Slow/old laptop?

I'd call it slightly overkill workstation. IPFS node sits on hdd bcache-d on an nvme ssd, using badger datastore with about 24G of data in it.

[lidel]

Interesting. Is there a specific IPFS Path that triggers the issue or is it purely random?

So far I've seen 'gui not being properly refreshed' once, when I was pinning root of a big directory tree that was not fully loaded into my node. Everything got back to normal after download finished, so I low-key suspect API stopped responding to new requests (for ipfs swarm peers or for ipfs pin ls <path>) when pinning of mentioned <path> was in progress. Does this sound familiar?

[magik6k]

ipfs swarm peers should be fast, ipfs pin ls may get really slow for mid-large repos, especially when called without -t recursive flag (mine repo has few 100k's of pins, pin ls can take minutes to finish)

[lidel]

Reopening, for some reason parameters added in #594 do not end up in the actual HTTP request sent by js-ipfs-api.