Then this is a kernel issue, I assumed that cifsd was part of smbclient and I see that taking up resources, but maybe that is unrelated (it was in a normal resource usage range).
Please reassign as you think appropriate. On 2017-03-05 14:36, Jelmer Vernooij wrote: > On Sun, Mar 05, 2017 at 12:30:33PM +0100, nils wrote: >> Package: smbclient >> Version: 2:4.2.14+dfsg-0+deb8u2 >> Severity: important >> >> Hello, >> after upgrading my kernel from 4.0.8 to 4.0.9 I am getting connection floods >> to >> my FreeNAS samba server from my debian machine. I never had this issue with >> 4.0.8 >> >> Linux dnet64 4.9.0-0.bpo.1-amd64 #1 SMP Debian 4.9.2-2~bpo8+1 (2017-01-26) >> x86_64 GNU/Linux >> ii smbclient 2:4.2.14+dfsg-0+deb8u2 amd64 command-line >> SMB/CIFS clients for Unix >> >> After mouting the cifs share(s) everything is OK for about 10 minutes, then a >> flood of connections happens. Here is an extract of 'netstat -an' run every >> second. Only one or two connections are shown for every 'netstat -an' run. >> The >> connections are opening and closing at an incredible speed (nethogs goes to >> 100% and doesn't show anything because it's overwhelmed). gkrellm reports >> about >> 70k/Sec traffic due to this. See how the source port numbers increase >> increadibly quickly to get an idea of how many connections are really >> happening... >> .... >> tcp 0 1 10.0.2.15:55044 192.168.2.88:445 SYN_SENT >> tcp 0 1 10.0.2.15:55252 192.168.2.88:445 SYN_SENT >> tcp 0 0 10.0.2.15:55288 192.168.2.88:445 >> ESTABLISHED >> tcp 0 1 10.0.2.15:55314 192.168.2.88:445 SYN_SENT >> tcp 0 1 10.0.2.15:55348 192.168.2.88:445 SYN_SENT >> tcp 0 0 10.0.2.15:55396 192.168.2.88:445 >> ESTABLISHED >> tcp 0 1 10.0.2.15:55454 192.168.2.88:445 SYN_SENT >> tcp 0 0 10.0.2.15:55500 192.168.2.88:445 >> ESTABLISHED >> tcp 0 0 10.0.2.15:55544 192.168.2.88:445 >> ESTABLISHED >> tcp 0 42 10.0.2.15:55586 192.168.2.88:445 >> ESTABLISHED >> tcp 0 0 10.0.2.15:55630 192.168.2.88:445 >> ESTABLISHED >> tcp 0 0 10.0.2.15:55676 192.168.2.88:445 >> ESTABLISHED >> tcp 0 0 10.0.2.15:55720 192.168.2.88:445 >> ESTABLISHED >> tcp 0 1 10.0.2.15:55770 192.168.2.88:445 SYN_SENT >> tcp 1 0 10.0.2.15:55820 192.168.2.88:445 >> CLOSE_WAIT >> tcp 0 1 10.0.2.15:55868 192.168.2.88:445 SYN_SENT >> tcp 0 0 10.0.2.15:55912 192.168.2.88:445 >> ESTABLISHED >> tcp 0 0 10.0.2.15:55962 192.168.2.88:445 >> ESTABLISHED >> tcp 1 0 10.0.2.15:56010 192.168.2.88:445 >> CLOSE_WAIT >> tcp 0 1 10.0.2.15:56058 192.168.2.88:445 SYN_SENT >> tcp 0 0 10.0.2.15:56056 192.168.2.88:445 LAST_ACK >> tcp 0 0 10.0.2.15:56104 192.168.2.88:445 >> ESTABLISHED >> tcp 0 0 10.0.2.15:56162 192.168.2.88:445 >> ESTABLISHED >> tcp 0 0 10.0.2.15:56202 192.168.2.88:445 >> ESTABLISHED >> tcp 0 1 10.0.2.15:56246 192.168.2.88:445 SYN_SENT >> tcp 0 0 10.0.2.15:56290 192.168.2.88:445 >> ESTABLISHED >> ....etc. >> >> Rebooting in 4.0.8 makes the issue go away, and I can have the shares mounted >> as long as I want without a connection flood. Hence I assume that it's a >> smbclient + Kernel 4.0.9 issue > smbclient is a command-line tool, if you're just mounting CIFS shares then > this > is an issue in the kernel rather than in smbclient. > > Jelmer
