You are not logged in.
After today's redis upgrade from (4.0.10-1 -> 4.0.11-1) unixsocket is no longer created. This is reproducible across multiple identical hosts. I looked over the Redis Wiki to see if something perhaps changed, but to no avail the documentation is the same.
From redis.conf
# unixsocket /tmp/redis.sock
unixsocket /run/redis/redis.sock
# unixsocketperm 700
unixsocketperm 770
Using systemd fails
# systemctl status -l redis
● redis.service - Advanced key-value store
Loaded: loaded (/etc/systemd/system/redis.service; enabled; vendor preset: disabled)
Active: failed (Result: exit-code) since Tue 2018-08-07 16:25:22 PDT; 2min 53s ago
Process: 471683 ExecStart=/usr/bin/redis-server /etc/redis.conf (code=exited, status=217/USER)
Main PID: 471683 (code=exited, status=217/USER)
Aug 07 16:25:22 venus systemd[1]: Started Advanced key-value store.
Aug 07 16:25:22 venus systemd[471683]: redis.service: Failed to determine user credentials: No such process
Aug 07 16:25:22 venus systemd[471683]: redis.service: Failed at step USER spawning /usr/bin/redis-server: No such process
Aug 07 16:25:22 venus systemd[1]: redis.service: Main process exited, code=exited, status=217/USER
Aug 07 16:25:22 venus systemd[1]: redis.service: Failed with result 'exit-code'
If the specified redis unix socket is manually created manually starting redis with /usr/bin/redis-server /etc/redis.conf starts up just fine.
For reference here is the distributed systemd service file which I have tried playing around with to no avail.
[Unit]
Description=Advanced key-value store
After=network.target
[Service]
Type=simple
User=redis
Group=redis
ExecStart=/usr/bin/redis-server /etc/redis.conf
ExecStop=/usr/bin/redis-cli shutdown
CapabilityBoundingSet=
PrivateTmp=true
PrivateDevices=true
ProtectSystem=full
ProtectHome=true
NoNewPrivileges=true
RuntimeDirectory=redis
RuntimeDirectoryMode=755
LimitNOFILE=10032
[Install]
WantedBy=multi-user.target
I remember not too long ago there being some systemd updates that basically were overzealous systemd service security options and was reported as a bug for other services using a unix socket.
Offline
Quick update - it looks like going from the recent systemd update 239.0-2 -> 238.76-1 fixes the problem. If anyone else can confirm the same problem with redis or any other service that uses a unix socket maybe we should open a systemd bug for systemd 239.0 unless it is confirmed a bug with the Arch specific 239.0-2 release.
Offline
I don't really understand why redis was not starting earlier today, but went ahead and reinstalled latest systemd 239.0-2, reloaded daemon and restarted redis just to verify startup - it did.
Offline
Same problem here today. Gitlab didn't come back up after a `pacman -Syu`, traced it to redis not starting, then to this. Forcing a re-install with `pacman -S systemd` did the trick.
Offline
Same problem with MariaDb today. Thanks to caleb, I figured that a deamon re-exec would solve this problem. (No reinstallation required)
systemctl daemon-reexec
Offline