Am Freitag, den 07.09.2018, 11:21 +0200 schrieb Dr. Bernd Feige: > Am Donnerstag, den 06.09.2018, 10:34 -0500 schrieb Steve French: > > Can you verify that /proc/fs/cifs/Stats (when the hang occurs) does > > not show additional session or share reconnects? > > No reconnects. I've got a bunch of files for you (attached). Maybe > some > general info: > cat /proc/version > Linux version 4.18.6-gentoo (root@pst15) (gcc version 8.2.0 (Gentoo > 8.2.0-r2 p1.2)) #1 SMP PREEMPT Thu Sep 6 11:53:00 CEST 2018 > cifs compiled as a module > > It really seems to be dependent upon RAM state (uninitialized > variable > or something?). After one warm reboot, I couldn't reproduce the > problem > with vers=2.1 or vers=3. After the next cold reboot, I could > reproduce > both with vers=2.1 and vers=3. BTW, I now found that the hangs don't > last infinitely, it just takes minutes. Any ideas? It's present in 4.18.7 as well. Versions up to 4.17.19 are rock solid. The Stats show some failed QueryDirectories; later, after the timeouts, failed IOCTLs appear. It's notable that just the process doing the listing is hung and can't be killed, with 100% CPU load; the same CIFS directory can be listed by another process which may succeed or also hang, and so on. I've also probed the AD server using nmap -O, it says: Warning: OSScan results may be unreliable because we could not find at least 1 open and 1 closed port Device type: general purpose Running: Microsoft Windows 2016 OS CPE: cpe:/o:microsoft:windows_server_2016 OS details: Microsoft Windows Server 2016 Thanks and best regards, Bernd
