ThinLinc 4.5.0 SLES12 SP3 (x86_64) Noticed this while upgrading from ThinLinc 4.4.0. I used tl-setup with an answer file. > 2017-12-15 10:59:27,123: Starting service 'tlwebadm'... > 2017-12-15 10:59:27,604: Traceback (most recent call last): > 2017-12-15 10:59:27,604: File "/opt/thinlinc/sbin/../libexec/tl-setup.py", line 566, in <module> > 2017-12-15 10:59:27,605: iIi1 ( ) > 2017-12-15 10:59:27,605: File "/opt/thinlinc/sbin/../libexec/tl-setup.py", line 523, in iIi1 > 2017-12-15 10:59:27,605: i11I1I1I = iIiii1iI1 . run ( ) > 2017-12-15 10:59:27,605: File "/opt/thinlinc/modules/thinlinc/wizard.py", line 100, in run > 2017-12-15 10:59:27,605: return self . _run_text ( ) > 2017-12-15 10:59:27,605: File "/opt/thinlinc/modules/thinlinc/wizard.py", line 118, in _run_text > 2017-12-15 10:59:27,605: oOOO00o ( ) > 2017-12-15 10:59:27,605: File "/opt/thinlinc/modules/thinlinc/tlsetup/services.py", line 176, in ii11i1 > 2017-12-15 10:59:27,605: OOO0OOO00oo = ii1 ( ) > 2017-12-15 10:59:27,606: File "/opt/thinlinc/modules/thinlinc/tlsetup/services.py", line 132, in ii1 > 2017-12-15 10:59:27,606: if not oO ( "tlwebadm" ) : > 2017-12-15 10:59:27,606: File "/opt/thinlinc/modules/thinlinc/tlsetup/services.py", line 86, in oO > 2017-12-15 10:59:27,606: ( o00ooooO0oO , oOoOo00oOo ) = iI11 . communicate ( ) > 2017-12-15 10:59:27,606: File "/opt/thinlinc/modules/subprocess.py", line 667, in communicate > 2017-12-15 10:59:27,606: return self._communicate(input) > 2017-12-15 10:59:27,606: File "/opt/thinlinc/modules/subprocess.py", line 1138, in _communicate > 2017-12-15 10:59:27,606: rlist, wlist, xlist = select.select(read_set, write_set, []) > 2017-12-15 10:59:27,606: error: (4, 'Interrupted system call') This happened only once. I was not able to reproduce the problem by re-running tl-setup.
This is an old ThinLinc version, but we haven't upgraded our copy of subprocess.py since then so it is likely the problem remains.
Upstream bug, fixed in subprocess.py shipped with Python 2.7: https://bugs.python.org/issue12493
Bug 4120 has been fixed now and we are now using the system version of subprocess. As such, we consider this an issue for the distribution to fix and not us. (most distributions will also be running a new enough python to have this fixed)