Occasionally, local ipfs node/daemon isn't restarted and running on browser restart

[stephendonner]

Description

Occasionally, local ipfs node/daemon isn't restarted and running on browser restart

Steps to Reproduce

1. load ipfs://bafybeiemxf5abjwjbikoz4mc3a3dla6ual3jsgpdr4cjr3oz3evfyavhwq/wiki/Vincent_van_Gogh.html in a new profile
2. click Use a local node on the interstitial page
3. load ipns://en.wikipedia-on-ipfs.org in a new tab
4. shut down Brave
5. re-launch the browser
6. wait for the reopened tabs to load
7. repeat steps 4-6 until you see the error below
8. load brave://ipfs-internals and look at the node status

Actual result:

At some point, you'll run into this Can't load resources using local node error; the daemon/local node isn't running, at this point

I don't see anything unusual in the logs:

generating ED25519 keypair...done
peer identity: 12D3KooWSffcx86kMLgkgaoT9iGATh3sqmSuiFtECkPKxQJXTJ42
initializing IPFS node at /Users/stephendonner/Library/Application Support/BraveSoftware/Brave-Browser-Nightly/brave_ipfs
to get started, enter:

	ipfs cat /ipfs/QmQPeNsJPyVWPFDVHb77w8G42Fvo15z4bG2X8D2GhfbSXc/readme

Error: daemon not running
Initializing daemon...
go-ipfs version: 0.7.0
Repo version: 10
System version: amd64/darwin
Golang version: go1.14.4
Swarm listening on /ip4/10.0.0.19/tcp/44002
Swarm listening on /ip4/127.0.0.1/tcp/44002
Swarm listening on /ip6/2002:6398:716c:0:1436:784e:aec9:6a76/tcp/44002
Swarm listening on /ip6/2002:6398:716c:0:2d0b:e962:90af:a5cf/tcp/44002
Swarm listening on /ip6/2002:6398:716c:0:5925:80a7:d4fa:eb29/tcp/44002
Swarm listening on /ip6/::1/tcp/44002
Swarm listening on /p2p-circuit
Swarm announcing /ip4/10.0.0.19/tcp/44002
Swarm announcing /ip4/127.0.0.1/tcp/44002
Swarm announcing /ip6/::1/tcp/44002
API server listening on /ip4/127.0.0.1/tcp/45002
WebUI: http://127.0.0.1:45002/webui
Gateway (readonly) server listening on /ip4/127.0.0.1/tcp/48081
Daemon is ready

Expected result:

Local ipfs node relaunches/restarts, and tab content loads upon browser restart/restore.

Reproduces how often:

It reproduces 100% of the time if you give the shutdown-and-restart steps enough tries; sometimes it takes a while for it to happen, and other times, it happens immediately on relaunch (less common). Perhaps it's a timing/race-condition thing.

Found while testing #14328.

Brave version (brave://version info)

Brave	1.24.4 Chromium: 89.0.4389.90 (Official Build) nightly (x86_64)
Revision	62eb262cdaae9ef819aadd778193781455ec7a49-refs/branch-heads/4389@{#1534}
OS	macOS Version 11.2.3 (Build 20D91)

@spylogsster @lidel any debug logging I can help add, here?

[spylogsster]

@stephendonner this looks like a problem I have already fixed here https://github.com/brave/brave-core/blame/master/components/services/ipfs/ipfs_service_impl.cc#L274, it should work in versions starting from https://github.com/brave/brave-core/releases/tag/v1.24.53 and I can not reproduce it on latest one, could you please recheck?

[stephendonner]

@stephendonner this looks like a problem I have already fixed here https://github.com/brave/brave-core/blame/master/components/services/ipfs/ipfs_service_impl.cc#L274, it should work in versions starting from https://github.com/brave/brave-core/releases/tag/v1.24.53 and I can not reproduce it on latest one, could you please recheck?

Cannot reproduce this using

Brave	1.24.52 Chromium: 90.0.4430.51 (Official Build) nightly (x86_64)
Revision	32e5fa33a31641bded70a90e60121060691e7125-refs/branch-heads/4430@{#927}
OS	macOS Version 11.2.3 (Build 20D91)