summary |
shortlog | log |
commit |
commitdiff |
tree
first ⋅ prev ⋅ next
Stephen Soltesz [Wed, 8 Aug 2007 13:36:46 +0000 (13:36 +0000)]
+ findbad.py: this actively probes all machines in the PLC db, using ping,
ssh, and then various commands on the machine to determine the actual bootstate.
These records are saved to disk for diagnose.py
+ diagnose.py: reads entries from findbad and previous actions, merging the
two together to determine if machines have improved, or gotten worse. All
actions to be performed are recorded and written to a diagnose_out pickle file
for action.py
+ action.py: reads the diagnose_out file from diagnose.py and performs the
actions. It permanently records the resuls in act_all pickle file.
These three in combination are Monitor.
Stephen Soltesz [Wed, 8 Aug 2007 13:32:43 +0000 (13:32 +0000)]
+ added 'production' namespace to the non-debug pickle files. This keeps
everything grouped together on a file list, and makes the mode very explicit.
Stephen Soltesz [Wed, 8 Aug 2007 13:31:32 +0000 (13:31 +0000)]
+ some cleanup. some dirtying.
Stephen Soltesz [Wed, 8 Aug 2007 13:30:42 +0000 (13:30 +0000)]
+ split the policy file into three classes: Merge(), Diagnose(), and Action().
This split is more natural and allows all the diagnosis/state-transition code to
live in once place. Action() is very simple, just taking the records from
Diagnose() and performing them.
Stephen Soltesz [Wed, 8 Aug 2007 13:28:55 +0000 (13:28 +0000)]
+ updated enableSliceCreation and enableSlices to reverse the effect of site
squeezing.
Stephen Soltesz [Wed, 8 Aug 2007 13:28:06 +0000 (13:28 +0000)]
+ add additional support for RT tickets, closing, changing Subject, and CCs.
emailViaRT() is the only needed call. If ticket_id is given, it uses
this, otherwise, a new ticket is created.
Stephen Soltesz [Wed, 8 Aug 2007 13:26:46 +0000 (13:26 +0000)]
+ add better messages for what to expect in the future
Stephen Soltesz [Wed, 8 Aug 2007 13:26:24 +0000 (13:26 +0000)]
+ format time record
Stephen Soltesz [Wed, 8 Aug 2007 13:25:57 +0000 (13:25 +0000)]
+ use OptionParser in optparse python module instead of getopt
Stephen Soltesz [Wed, 8 Aug 2007 13:25:11 +0000 (13:25 +0000)]
+ allow None arguments to constructor, and generate good defaults
Stephen Soltesz [Mon, 30 Jul 2007 13:51:20 +0000 (13:51 +0000)]
shouldn't be in cvs.
Stephen Soltesz [Tue, 3 Jul 2007 19:59:02 +0000 (19:59 +0000)]
+ new XMLRPC_SERVER name to boot.planet-lab.org
Stephen Soltesz [Tue, 3 Jul 2007 19:58:34 +0000 (19:58 +0000)]
+ use the emailViaRT() for email rather than standard email
Stephen Soltesz [Tue, 3 Jul 2007 19:57:59 +0000 (19:57 +0000)]
+ added temporary fix for ignoring tickets with a blacklist
Stephen Soltesz [Tue, 3 Jul 2007 19:57:16 +0000 (19:57 +0000)]
+ introduced rt command line emailViaRT
Stephen Soltesz [Tue, 3 Jul 2007 19:56:45 +0000 (19:56 +0000)]
+ updated tech guide url to go directly to NodeInstallation
Stephen Soltesz [Fri, 29 Jun 2007 12:42:22 +0000 (12:42 +0000)]
+ monitor.py -- modified the following three to use a record-based events,
rather than node-based
+ comon.py -- currently only looks at dbg nodes.
+ policy.py -- separated diagnoseSite() from actOnSite()
+ rt.py -- Retrieve all tickets once
+ config.py -- store for command line arguments used by other utilities.
Awkward.
+ emailTxt.py -- new messages for escalation.
+ mailer.py -- added a bcc option and hooks for config() options
+ plc.py -- added a few extra fields and utility functions
Stephen Soltesz [Fri, 29 Jun 2007 12:38:36 +0000 (12:38 +0000)]
+ blacklist.py -- manages a node blacklist on which no actions should ever be
taken
+ bootcds.py -- collects bootcd information from debug state nodes
+ bwlimit.py -- fetch all nodes with broken bwlimits.
+ dumpact.py -- pretty print the act_all.pkl db generated by monitor.py
+ getnodekey.py -- generate a known_hosts file based on the ssh_rsa_key field
of the PLC node db.
+ printpdb.py -- another pretty printer for pickle files.
+ soltesz.py -- utilitiy functions for pickles, config, etc.
Stephen Soltesz [Fri, 29 Jun 2007 12:32:58 +0000 (12:32 +0000)]
- I don't know how these ended up in cvs.
Faiyaz Ahmed [Wed, 16 May 2007 01:53:46 +0000 (01:53 +0000)]
Rewrite of policy engine.
Marc Fiuczynski [Thu, 19 Apr 2007 20:43:00 +0000 (20:43 +0000)]
added 'cleanSlices' to remove disabled users from a slice
Faiyaz Ahmed [Fri, 6 Apr 2007 17:38:14 +0000 (17:38 +0000)]
Increase threshold to a week for slice creation, 2 weeks for suspension.
Faiyaz Ahmed [Fri, 6 Apr 2007 16:16:54 +0000 (16:16 +0000)]
Update to new API.
Faiyaz Ahmed [Mon, 2 Apr 2007 20:59:37 +0000 (20:59 +0000)]
plctool - Marc's CLI util.
config.py - debug=false
Faiyaz Ahmed [Mon, 2 Apr 2007 20:57:57 +0000 (20:57 +0000)]
Migrate to new API.
Faiyaz Ahmed [Mon, 2 Apr 2007 20:28:50 +0000 (20:28 +0000)]
Fixed syntax error in passing PCU info.
Marc Fiuczynski [Thu, 22 Feb 2007 17:09:33 +0000 (17:09 +0000)]
- set API URL to www.planet-lab.org
- add authCheck method
Marc Fiuczynski [Mon, 19 Feb 2007 17:42:21 +0000 (17:42 +0000)]
fleshed out slice enable/disable support
Marc Fiuczynski [Mon, 12 Feb 2007 19:59:00 +0000 (19:59 +0000)]
Replace enableSliceCreation/removeSliceCreation with a single setSliceMax function.
Marc Fiuczynski [Mon, 12 Feb 2007 19:54:56 +0000 (19:54 +0000)]
o Fixed removeSliceCreation and enableSliceCreation functions to work with
new API.
Marc Fiuczynski [Mon, 12 Feb 2007 19:15:08 +0000 (19:15 +0000)]
o Fixed slices() function to use new API.
Marc Fiuczynski [Thu, 8 Feb 2007 22:43:11 +0000 (22:43 +0000)]
updated a number of functions to use new API
Marc Fiuczynski [Thu, 8 Feb 2007 19:59:03 +0000 (19:59 +0000)]
- Fix siteId() to work with new API
Marc Fiuczynski [Thu, 8 Feb 2007 19:43:09 +0000 (19:43 +0000)]
- Fix nodesDbg to use GetNodes API because Anon* API is now gone.
- Add renewAllSlices function to move forward all slice expiration dates that
are sooner than the date given as an argument.
- Add "allow_none=True" argument to xmlrpclib.Server so that None arg can be
marshalled via API.
Marc Fiuczynski [Thu, 1 Feb 2007 14:25:56 +0000 (14:25 +0000)]
check if maxslices arg is pased to enableSliceCreation
Marc Fiuczynski [Thu, 1 Feb 2007 14:20:19 +0000 (14:20 +0000)]
check if maxslices arg is pased to enableSliceCreation
Marc Fiuczynski [Wed, 24 Jan 2007 19:29:44 +0000 (19:29 +0000)]
updated so that plc.py can be used also nicely from the command line
Faiyaz Ahmed [Wed, 17 Jan 2007 19:46:40 +0000 (19:46 +0000)]
Update log.
Faiyaz Ahmed [Wed, 17 Jan 2007 19:33:04 +0000 (19:33 +0000)]
Added check so we dont keep resending the same email
Faiyaz Ahmed [Wed, 17 Jan 2007 16:03:30 +0000 (16:03 +0000)]
Except on MTA error and continue.
Faiyaz Ahmed [Thu, 11 Jan 2007 21:39:07 +0000 (21:39 +0000)]
Changed time to act on emails. WE GET TOO MUCH EMAIL ALREADY!
Faiyaz Ahmed [Wed, 10 Jan 2007 20:08:44 +0000 (20:08 +0000)]
* Emails users when slice renewal/creation is suspended, and when their slices are suspended.
Faiyaz Ahmed [Wed, 10 Jan 2007 20:06:30 +0000 (20:06 +0000)]
Hosed this file on alfred when rins'ing. Its RO DB access so (hopefullY) doesn't pose a serious lack of s3curity.
* Contains auth info for RT.
Faiyaz Ahmed [Tue, 14 Nov 2006 19:38:34 +0000 (19:38 +0000)]
*** empty log message ***
Faiyaz Ahmed [Tue, 14 Nov 2006 19:36:09 +0000 (19:36 +0000)]
SSH and telnet library
Faiyaz Ahmed [Tue, 14 Nov 2006 19:27:09 +0000 (19:27 +0000)]
*** empty log message ***
Faiyaz Ahmed [Tue, 14 Nov 2006 19:20:13 +0000 (19:20 +0000)]
* Sets nodes to reboot, uses PCU if available. Defaults to POD/email (with site squeezing)
* Slice emails, site slice creation revoke, freeze running slices
* Changed mailto target for summary email
Faiyaz Ahmed [Fri, 27 Oct 2006 20:24:24 +0000 (20:24 +0000)]
* Emails PI, then Slices if the node does not come up after a certain number of days.
* Beginnings of slice freeze and node rins via PLC api. Still need to finish PCU stuff.
Faiyaz Ahmed [Tue, 24 Oct 2006 20:27:32 +0000 (20:27 +0000)]
Uses CoMon's ability to find 'upness' to email. Changed queueing between threads; no more huge sleeps to maintain synch.
* Individual email messages per bucket
* emailed{} now pickled so as not to email the same site more than once.
* PLC anon API to get site basenames for email
* Searches RT for open/new in support or offline for tickets with hostname
* Beginnings of real chek for dbg
* Started squeezing of slices via PLC api.
* Work in progress. Not ready for human consumption.
Faiyaz Ahmed [Tue, 24 Oct 2006 20:19:06 +0000 (20:19 +0000)]
*** empty log message ***
Faiyaz Ahmed [Tue, 3 Oct 2006 21:45:59 +0000 (21:45 +0000)]
This commit was generated by cvs2svn to compensate for changes in r2,
which included commits to RCS files with non-trunk default branches.
Planet-Lab Support [Tue, 3 Oct 2006 21:45:59 +0000 (21:45 +0000)]
New repository initialized by cvs2svn.