Brennan freeze. Large number of smbd zombie processes?

[Jon Degenhardt]

My Brennan B2 became unresponsive after being up and running for about a week. I had been using it off an on during that period. It is setup as a NAS server accessed by Sonos speakers (gen 2). It had been played both starting from the Brennan and from Sonos. At one point it refused connections to the music folder exported as NAS. Couldn't reach it from either Sonos or MacOS. Front panel worked though. Indicated it was compressing (0 of 16), but there was nothing to compress. I ran the "reboot" command from the "maintenance" menu. It stayed in "rebooting" mode, never exiting.

After about 20 minutes I ssh'ed to the brennan to see if there was anything obvious going on (it did accept ssh connections). I ran the 'ps' command and saw a very large number of zombie smbd (SMB daemon) processes. Didn't check the number, but a very long list.

At this point I power cycled the brennan and it came back and ran normally. I don't know if the zombie smbd processes were responsible for the unresponsive behavior. Could be many things. But it's something that might be worth Brennan tech support checking for.

I did some follow-up checks, and it does appear that zombie smbd processes gradually grow without being reaped. They don't appear to be consuming any cpu resources, but they do consume an entry in the process table. The kernel.threads-max value is only 6211, so it is possible the process table became filled.

    # sysctl kernel.threads-max

    kernel.threads-max = 6211

As an example, here are a few lines from the middle of a ps command. The square brackets indicate a zombie process.

   # ps | head -n 500 | tail -n 5
    2216 root     [smbd]
    2218 root     [smbd]
    2219 root     [smbd]
    2221 root     [smbd]
    2253 root     [smbd]

To check growth I ran the ps command every 2 minutes after rebooting and playing some music (from Sonos). The command below prints the count every two minutes, so this is the first 30 minutes:

   # for i in $(seq 1 200) ; do echo -n "$i - " ; ps | grep '\[smbd\]' | wc -l; sleep 120; done
   1 - 0
   2 - 2
   3 - 34
   4 - 47
   5 - 90
   6 - 131
   7 - 136
   8 - 179
   9 - 194
   10 - 202
   11 - 202
   12 - 220
   13 - 220
   14 - 220
   15 - 237

The growth was quick initially, then slowed. When nothing is playing there is about 1 new zombie smbd process every 15 minutes. Growth is faster when playing music.

Again. I don't know if these zombie processes are causing a problem. Could be that they will eventually get cleaned up, and they are unrelated to the unresponsive behavior I saw. But something that might be worth checking in to if its not already understood.

--Jon

PS. Have had my B2 for about a month. Very happy with the B2!

[Daniel Taylor]

I'm not an expert - far from it.  What the growing number of SMB processes suggest to me is the possibility that your B2's WiFi signal strength is not all it should be.  If the connection were momentarily dropped, it could cause a new SMB process to be spawned.  The bug would be that the old processes are left to clutter things up and consume memory.

Standard WiFi Signal Strength Advise:

Check your WiFi signal strength using Settings > Maintenance > WiFi Strength.  (I got that from the Brennan website on the Menu page.)
Your signal strength should be better than -50 (-40 being better, -60 being worse) and not fluctuate wildly.
To improve your signal strength, try the following:
Move your B2 closer to your router.
Switch the WiFi dongle to USB C.
Move your WiFi dongle to the end of a USB extension cable.
You can buy a new WiFi dongle that has an antenna.  Look for one that has the RT5370 chipset.
If none of that helps, you could try plugging a network cable between the inside back of the B2 and your router (and remove the WiFi dongle).

[Jon Degenhardt]

Hi Daniel,

Thanks for the tip about wifi strength. I was unaware of this. It does not appear likely to be the issue. Wifi strength varies between -9 and -19 dBm, which is pretty good. And, the regularity of the change in smbd process growth indicates it's a not a result of random disconnects. For example, the 1 new process every 15 minutes repeats predictably for many hours. (I'm guessing that Sonos makes scheduled connections, but don't know for sure.) Growth also appears related to song boundaries, though I didn't assemble enough data to be certain. That is, the count appears to go up when Sonos switches to a new song, but not during a song.

--Jon

[Daniel Taylor]

That's interesting data that you've gathered.  It troubles me that the SMB process count keeps going up.  I don't see how that can continue indefinitely without causing a problem.   I hope the folks at Brennan will take a look.

[PMB]

Hi Jon,

Thank you for the info which I have forwarded to Martin Brennan.

Paul

Brennan Support.

[Jon Degenhardt]

Good chance this problem arises from Sonos behavior and largely addressed in the recent Sonos 14.4 S2 update. See: https://en.community.sonos.com/music-services-and-sources-228994/sonos-making-thousands-of-smb3-connections-to-synology-6864620.

After updating to the Sonos 14.4 release there is still growth in zombie SMBD and NMBD processes, but the rate has clearly slowed. The rate will become clearer after the setup has been running a few days.

--Jon