Bug 1203900
Summary: | ansible failing file copies to rawhide machine because of sshd error: mm_answer_audit_end_command: invalid handle | ||
---|---|---|---|
Product: | [Fedora] Fedora | Reporter: | Josh Boyer <jwboyer> |
Component: | openssh | Assignee: | Jakub Jelen <jjelen> |
Status: | CLOSED ERRATA | QA Contact: | Fedora Extras Quality Assurance <extras-qa> |
Severity: | unspecified | Docs Contact: | |
Priority: | unspecified | ||
Version: | 22 | CC: | a.badger, athmanem, bobi_relv, jjelen, jwboyer, kevin, kupo, mattias.ellert, maxim, mgrepl, plautrba, tmraz |
Target Milestone: | --- | Keywords: | Reopened |
Target Release: | --- | ||
Hardware: | Unspecified | ||
OS: | Unspecified | ||
Whiteboard: | |||
Fixed In Version: | openssh-6.8p1-4.fc22 | Doc Type: | Bug Fix |
Doc Text: | Story Points: | --- | |
Clone Of: | Environment: | ||
Last Closed: | 2015-04-21 19:06:14 UTC | Type: | Bug |
Regression: | --- | Mount Type: | --- |
Documentation: | --- | CRM: | |
Verified Versions: | Category: | --- | |
oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |
Cloudforms Team: | --- | Target Upstream Version: | |
Embargoed: |
Description
Josh Boyer
2015-03-19 22:10:44 UTC
Seems to be ssh: openssh-6.7p1-11.fc23.x86_64 - fails openssh-6.7p1-10.fc23.x86_64 - works fine when -11 is installed I see in audit.log: type=ANOM_ABEND msg=audit(1426806655.424:7732): auid=100037 uid=0 gid=0 ses=54 subj=system_u:system_r:sshd_t:s0-s0:c0.c1023 pid=25840 comm="sshd" exe="/usr/sbin/sshd" sig=11 type=ANOM_ABEND msg=audit(1426806658.465:7768): auid=100037 uid=0 gid=0 ses=55 subj=system_u:system_r:sshd_t:s0-s0:c0.c1023 pid=25871 comm="sshd" exe="/usr/sbin/sshd" sig=11 ie, segfault? Happy to gather more info, etc... Thank you for reporting this issue. It was introduced by wrong applying a downstream patch related to auditing in recent version. Sorry for the inconvenience. I prioritize this problem after finishing current rebase, but I wanted to let you know that it was picked up and something is going on. Thanks to both of you. I also downgraded to -10 and ansible is working again. I'll be happy to test when a fix is available. *** Bug 1204494 has been marked as a duplicate of this bug. *** This issue should be resolved with new openssh-6.8p1-1.1 release in rawhide, which contains this bug fixed: http://koji.fedoraproject.org/koji/buildinfo?buildID=623115 If it will not solve your problem, please comment. openssh-6.8p1-2.fc22 has been submitted as an update for Fedora 22. https://admin.fedoraproject.org/updates/openssh-6.8p1-2.fc22 Package openssh-6.8p1-2.fc22: * should fix your issue, * was pushed to the Fedora 22 testing repository, * should be available at your local mirror within two days. Update it with: # su -c 'yum update --enablerepo=updates-testing openssh-6.8p1-2.fc22' as soon as you are able to. Please go to the following url: https://admin.fedoraproject.org/updates/FEDORA-2015-4898/openssh-6.8p1-2.fc22 then log in and leave karma (feedback). openssh-6.8p1-2.fc22 has been pushed to the Fedora 22 stable repository. If problems still persist, please make note of it in this bug report. This is still failing with openssh-6.8p1-2.fc22. Ansible file copies continue to fail and I can see the following in the journal: Apr 1 15:02:45 obiwan sshd[1982]: fatal: mm_answer_audit_end_command: invalid handle Apr 1 15:02:45 obiwan sshd[1982]: pam_unix(sshd:session): session closed for user jwboyer Apr 1 15:02:46 obiwan sshd[2064]: fatal: mm_request_send: write: Broken pipe Apr 1 15:02:46 obiwan sshd[2064]: fatal: mm_request_send: write: Broken pipe Apr 1 20:12:46 obiwan sshd[2643]: Accepted publickey for jwboyer from 192.168.11.63 port 55077 ssh2: RSA SHA256:raEyfyre5HGK2Ck3mDGU8iL8iXbNBtwbhFTep7tEiBE Apr 1 20:12:46 obiwan systemd: pam_unix(systemd-user:session): session opened for user jwboyer by (uid=0) Hi, I was playing around Ansible, but I still can't reproduce your issue through Ansible. Using latest rawhide version openssh-6.8p1-3.fc23 Using command from closed duplicate bug I don't get any error > ansible -i hosts all -m ping -vvvv I went again through audit and fixed some cases better way. By hand it seems working for me. Can you test it again with this build and report back? http://koji.fedoraproject.org/koji/taskinfo?taskID=9398779 If I got this right, it is problem of ssh commands with "ControlPersist option", which is commonly used by Ansible and persistent master is failing. Sorry for inconvenience. A simple file copy with openssh-6.7p1-10.fc23.x86_64 on the target machine: [jwboyer@vader kernel]$ ansible nuc-i7 -m copy -a "src=x86_64/kernel-4.0.0-0.rc4.git1.3.fc23.x86_64.rpm dest=~/." nuc-i7 | success >> { "changed": true, "checksum": "07016b5a3cb4e156210e31b163098ead3c9c4eab", "dest": "/home/jwboyer/./kernel-4.0.0-0.rc4.git1.3.fc23.x86_64.rpm", "gid": 1000, "group": "jwboyer", "md5sum": "aa40d75c6856bdc93c0c501a34e84b93", "mode": "0664", "owner": "jwboyer", "secontext": "unconfined_u:object_r:user_home_t:s0", "size": 60664, "src": "/home/jwboyer/.ansible/tmp/ansible-tmp-1427973730.47-175909122124082/source", "state": "file", "uid": 1000 } [jwboyer@vader kernel]$ The same file copy to the same machine with openssh-6.8p1-3.fc23.x86_64: [jwboyer@vader kernel]$ ansible nuc-i7 -m copy -a "src=x86_64/kernel-4.0.0-0.rc4.git1.3.fc23.x86_64.rpm dest=~/." nuc-i7 | FAILED => failed to transfer file to /home/jwboyer/.ansible/tmp/ansible-tmp-1427973906.05-222325072018028/source: Couldn't read packet: Connection reset by peer [jwboyer@vader kernel]$ The ping command also fails. [jwboyer@vader kernel]$ ansible nuc-i7 -m ping -vvvv <nuc-i7> <nuc-i7> <nuc-i7> ConnectTimeout=10 PasswordAuthentication=no KbdInteractiveAuthentication=no ControlPath=/home/jwboyer/.ansible/cp/ansible-ssh-%h-%p-%r PreferredAuthentications=gssapi-with-mic,gssapi-keyex,hostbased,publickey ControlMaster=auto Port=22 ControlPersist=60s <nuc-i7> nuc-i7 | FAILED => failed to transfer file to /home/jwboyer/.ansible/tmp/ansible-tmp-1427973970.96-163092021662184/ping: Couldn't read packet: Connection reset by peer I can test with the f22 build if you'd like, but I think the rawhide version of it shows the same issues. (In reply to Jakub Jelen from comment #10) > Hi, > I was playing around Ansible, but I still can't reproduce your issue through > Ansible. Using latest rawhide version openssh-6.8p1-3.fc23 > > Using command from closed duplicate bug I don't get any error > > ansible -i hosts all -m ping -vvvv > > > I went again through audit and fixed some cases better way. By hand it seems > working for me. Can you test it again with this build and report back? > http://koji.fedoraproject.org/koji/taskinfo?taskID=9398779 > > If I got this right, it is problem of ssh commands with "ControlPersist > option", which is commonly used by Ansible and persistent master is failing. > Sorry for inconvenience. OK, I tested that build on an f22 machine with successful results: [jwboyer@vader kernel]$ ansible obiwan -m copy -a "src=x86_64/kernel-4.0.0-0.rc4.git1.3.fc23.x86_64.rpm dest=~/." obiwan | success >> { "changed": true, "checksum": "07016b5a3cb4e156210e31b163098ead3c9c4eab", "dest": "/home/jwboyer/./kernel-4.0.0-0.rc4.git1.3.fc23.x86_64.rpm", "gid": 0, "group": "root", "md5sum": "aa40d75c6856bdc93c0c501a34e84b93", "mode": "0644", "owner": "jwboyer", "secontext": "unconfined_u:object_r:user_home_t:s0", "size": 60664, "src": "/home/jwboyer/.ansible/tmp/ansible-tmp-1427976697.45-116497879441197/source", "state": "file", "uid": 1000 } [jwboyer@vader kernel]$ ansible obiwan -m ping -vvvv<obiwan> <obiwan> <obiwan> ConnectTimeout=10 PasswordAuthentication=no KbdInteractiveAuthentication=no ControlPath=/home/jwboyer/.ansible/cp/ansible-ssh-%h-%p-%r PreferredAuthentications=gssapi-with-mic,gssapi-keyex,hostbased,publickey ControlMaster=auto Port=22 ControlPersist=60s <obiwan> <obiwan> ConnectTimeout=10 PasswordAuthentication=no 'LANG=C LC_CTYPE=C /usr/bin/python /home/jwboyer/.ansible/tmp/ansible-tmp-1427976707.55-205929320515257/ping; rm -rf /home/jwboyer/.ansible/tmp/ansible-tmp-1427976707.55-205929320515257/ >/dev/null 2>&1' KbdInteractiveAuthentication=no ControlPath=/home/jwboyer/.ansible/cp/ansible-ssh-%h-%p-%r PreferredAuthentications=gssapi-with-mic,gssapi-keyex,hostbased,publickey ControlMaster=auto Port=22 ControlPersist=60s obiwan | success >> { "changed": false, "ping": "pong" } [jwboyer@vader kernel]$ [jwboyer@obiwan ~]$ rpm -q openssh openssh-6.8p1-3.fc22.x86_64 [jwboyer@obiwan ~]$ Same build tested on the previously tested rawhide machine works too: [jwboyer@vader kernel]$ ansible nuc-i7 -m copy -a "src=x86_64/kernel-4.0.0-0.rc4.git1.3.fc23.x86_64.rpm dest=~/." nuc-i7 | success >> { "changed": true, "checksum": "07016b5a3cb4e156210e31b163098ead3c9c4eab", "dest": "/home/jwboyer/./kernel-4.0.0-0.rc4.git1.3.fc23.x86_64.rpm", "gid": 1000, "group": "jwboyer", "md5sum": "aa40d75c6856bdc93c0c501a34e84b93", "mode": "0664", "owner": "jwboyer", "secontext": "unconfined_u:object_r:user_home_t:s0", "size": 60664, "src": "/home/jwboyer/.ansible/tmp/ansible-tmp-1427976857.54-175393144647249/source", "state": "file", "uid": 1000 } [jwboyer@vader kernel]$ ansible nuc-i7 -m ping -vvvv<nuc-i7> <nuc-i7> <nuc-i7> ConnectTimeout=10 PasswordAuthentication=no KbdInteractiveAuthentication=no ControlPath=/home/jwboyer/.ansible/cp/ansible-ssh-%h-%p-%r PreferredAuthentications=gssapi-with-mic,gssapi-keyex,hostbased,publickey ControlMaster=auto Port=22 ControlPersist=60s <nuc-i7> <nuc-i7> ConnectTimeout=10 PasswordAuthentication=no 'LANG=C LC_CTYPE=C /usr/bin/python /home/jwboyer/.ansible/tmp/ansible-tmp-1427976861.58-200531392598203/ping; rm -rf /home/jwboyer/.ansible/tmp/ansible-tmp-1427976861.58-200531392598203/ >/dev/null 2>&1' KbdInteractiveAuthentication=no ControlPath=/home/jwboyer/.ansible/cp/ansible-ssh-%h-%p-%r PreferredAuthentications=gssapi-with-mic,gssapi-keyex,hostbased,publickey ControlMaster=auto Port=22 ControlPersist=60s nuc-i7 | success >> { "changed": false, "ping": "pong" } [jwboyer@vader kernel]$ Josh, thanks for the help with investigation. It appeared that I was unable to reproduce with root user, but non-root was failing as stated above. Update is on its way. openssh-6.8p1-4.fc22 has been submitted as an update for Fedora 22. https://admin.fedoraproject.org/updates/openssh-6.8p1-4.fc22 Package openssh-6.8p1-4.fc22: * should fix your issue, * was pushed to the Fedora 22 testing repository, * should be available at your local mirror within two days. Update it with: # su -c 'yum update --enablerepo=updates-testing openssh-6.8p1-4.fc22' as soon as you are able to. Please go to the following url: https://admin.fedoraproject.org/updates/FEDORA-2015-5526/openssh-6.8p1-4.fc22 then log in and leave karma (feedback). Hi Josh, would you consider this as a blocker for Beta release (currently postponed so we have time for this) or is it fine for you to have it in GA? I'm not sure if there are some test matrices for this Ansible use case for Beta. I'm not aware of any tests that would cause this to be a Beta blocker. Given that normal SSH sessions work fine and it is limited to a small set of file copies, I think just having it in GA should work. That's just my opinion though. If you'd like it to be blocker, then I'm fine with that too. openssh-6.8p1-4.fc22 has been pushed to the Fedora 22 stable repository. If problems still persist, please make note of it in this bug report. |