sdwdate sclockadj ruby-inline require bug

[phabricator-migrator]

Information

ID: 30
PHID: PHID-TASK-a3hyvlrc7qmnhr45bh7y
Author: Patrick
Status at Migration Time: resolved
Priority at Migration Time: Normal

Description

5767: Launching into background:       sudo          INLINEDIR=/var/cache/sdwdate             /usr/lib/sdwdate/sclockadj                --no-verbose                --no-debug                --no-first-wait                                --move-min 500000                --move-max 500000                --wait-min 1000000000                --wait-max 1000000000                --subtract 1290159494    
5767: Started subshell for sclockadj with pid: 25369
5767: sdwdate_subshell_read: start
5767: dispatching post_success (SDW_MODE: daemon): /usr/lib/timesync/timesync_post_success --autostart --identifier "timesync" --progressbaridx "$ID" --mode "$SDW_MODE" --whoami "$who_ami"
5767: sdwdate_subshell_read: sclockadj reports: Time before running sclockadj: 1418677914.833885947 [Mon Dec 15 21:11:54 UTC 2014]
5767: sdwdate_subshell_read: sclockadj reports: sdwdate_subshell_wait: Waiting for pid to finish SDWDATE_SCLOCKADJ_COMMAND_PID: 25380
5767: sdwdate_subshell_read: sclockadj reports: /usr/lib/ruby/1.9.1/rubygems/custom_require.rb:36:in `require': cannot load such file -- /var/cache/sdwdate/.ruby_inline/ruby-1.9.1/Inline_Cinline_45aeaf467303ff06d9c25b91daa74e32.so (LoadError)
5767: sdwdate_subshell_read: sclockadj reports: from /usr/lib/ruby/1.9.1/rubygems/custom_require.rb:36:in `require'
5767: sdwdate_subshell_read: sclockadj reports: from /usr/lib/ruby/vendor_ruby/inline.rb:531:in `load'
5767: sdwdate_subshell_read: sclockadj reports: from /usr/lib/ruby/vendor_ruby/inline.rb:844:in `inline'
5767: sdwdate_subshell_read: sclockadj reports: from /usr/lib/sdwdate/sclockadj:189:in `<class:Cinline>'
5767: sdwdate_subshell_read: sclockadj reports: from /usr/lib/sdwdate/sclockadj:187:in `<main>'
5767: sdwdate_subshell_read: sclockadj reports: sdwdate_subshell_wait: SDWDATE_SCLOCKADJ_COMMAND_PID: 25380 | SDWDATE_SCLOCKADJ_COMMAND_EXIT_CODE: 1
5767: sdwdate_subshell_read: sclockadj reports: was running for 0 s [~ 0.00 min] [~ 0.00 h].
5767: sdwdate_subshell_read: sclockadj reports: Time after running sclockadj: 1418677914.953156268 [Mon Dec 15 21:11:54 UTC 2014]
5767: sdwdate_subshell_read: sclockadj reports: exit code: 1
5767: sdwdate_subshell_read: end

Spontaneously happened out of nowhere. Worked before.

Could not reproduce this by running

sudo          INLINEDIR=/var/cache/sdwdate             /usr/lib/sdwdate/sclockadj                --no-verbose                --no-debug                --no-first-wait                                --move-min 500000                --move-max 500000                --wait-min 1000000000                --wait-max 1000000000                --subtract 1290159494

manually afterwards.

Any idea why this could happen?

Comments

---

Patrick

2014-12-16 01:09:55 UTC

By the way, this also happed before changing INLINEDIR once.

That’s what originally was motivated me to do this:

• sclockadj indent fix #369
• sclockadj does not fail when run without right to change clock / clock_(gettime/settime) return code checking #370
• sclockadj home folder permission issue #365

Seems to happen only seldom. But happens apparently.

---

Patrick

2014-12-16 01:12:03 UTC

Probably unrelated: sclockadj stuck at 100% CPU.

---

Patrick

2014-12-16 01:34:18 UTC

After some searching, I conclude, it could be related to permissions:
sdwdate/debian/sdwdate.postinst at master · Kicksecure/sdwdate · GitHub
Would perhaps be better if owned by root.

---

Patrick

2014-12-16 01:44:05 UTC

Attempt to fix:
permission fix · Kicksecure/sdwdate@ddab450 · GitHub

---

Patrick

2014-12-16 01:58:25 UTC

Added more debugging info:
debugging · Kicksecure/sdwdate@2b76871 · GitHub

---

Patrick

2015-01-15 17:24:22 UTC

Any idea @JasonJAyalaP?

---

JasonJAyalaP

2015-01-16 19:22:31 UTC

It looks like ruby_inline thinks there’s pre-compiled code in the inline dir, tries to load it, but fails. I don’t know if that’s because there isn’t the compiled code (and ruby inline should have compiled it) or some other reason. That load error is too generic to be sure.

---

Patrick

2015-01-16 19:30:11 UTC

Ok. Closed for now. Should be fixed. Didn’t run into this issue for a while now. And if it happens again, there will be helpful debug output. In case I notice again, I’ll reopen.