More information. - Curator version seems not changed and the version is 2.4.2. - The team is using InterProcessMutex recipe. - "not always deleted" means the znode isn't deleted
Here is the info for the most recent lock not being released: *Path* /Jabba/agent/pop/-750322352/_c_07036068-c848-4720-b553-d7b446d413a3-lock-0000001573 *Stat*czxid: 730803957058, mzxid: 730803957058, ctime: 1401067656032, mtime: 1401067656032, version: 0, cversion: 0, aversion: 0, ephemeralOwner: 4487297975270022858, dataLength: 16, numChildren: 0, pzxid: 730803957058*Data Bytes*6a 61 62 62 61 2d 69 2d 32 38 65 39 38 38 37 38*Data as String* jabba-i-<instanceID> On Tue, May 27, 2014 at 6:35 AM, Jordan Zimmerman < [email protected]> wrote: > Somehow these locks are being created but when cleared out by the finally, > they seem to not always be deleted. > > What do you mean by “not always be deleted”? Do you mean the lock isn’t > released? How do you know? > > Also, Is this in 2.4.2? Which recipe? InterProcessMutex? Nothing has > changed in that code that I know of (though it uses code that has changed > for other reasons). There are numerous tests for locks and they all > currently pass. Any more details on this? > > -JZ > > From: Bae, Jae Hyeon [email protected] > Reply: [email protected] [email protected] > Date: May 27, 2014 at 1:10:48 AM > To: [email protected] [email protected] > Subject: Issue with mutexes not being released > > Hi > > I got reported mutexes not being released. Somehow these locks are being > created but when cleared out by the finally, they seem to not always be > deleted. This just suddenly started happening since they upgraded curator > version, despite there being no changes to this code at all for quite > some time. > > Do you have any idea what have changed? > > Thank you > Best, Jae > >
