ClearFoundation Tracker - ClearOS
View Issue Details
0001562ClearOSapp-base - Base Systempublic2014-02-13 09:452015-03-25 08:02
user2 
user2 
normalminorN/A
closedfixed 
 
6.6.0 Beta 26.6.0 Beta 2 
0001562: Yum cache is still problematic
We still see reports of yum caching extremely old data. For example, a customer recently reported a cache that contained data that was at least 4 days old! In 6.5.0, the following parameter was added to yum.conf:

http_caching=packages

There's also the "metadata_expire" (defaults to 6 hours) and "mirrorlist_expire" parameters that might need to be shortened. That still doesn't explain the "4 days old" behavior. Perhaps it's related to skewed clocks at install time?
No tags attached.
has duplicate 0001524closed user2 Yum cache is still problematic 
related to 0001728closed user2 Yum cache is still problematic 
related to 0002279closed user2 yum upgrade can somtimes fail 
Issue History
2014-02-13 09:45user2New Issue
2014-02-13 09:45user2Issue generated from: 0001524
2014-02-13 09:45user2Relationship addedhas duplicate 0001524
2014-02-13 09:45user2Statusnew => confirmed
2014-04-30 05:27user2Target Version6.5.0 Updates => 6.6.0 Beta 2
2014-05-30 14:12user2Categoryapp-software-updates - Software Updates => app-base - Base System
2014-05-30 14:19user2Note Added: 0001193
2014-05-30 14:23user2Issue cloned: 0001728
2014-05-30 14:23user2Relationship addedrelated to 0001728
2014-05-30 14:45user2Statusconfirmed => resolved
2014-05-30 14:45user2Fixed in Version => 6.6.0 Beta 2
2014-05-30 14:45user2Resolutionopen => fixed
2014-05-30 14:45user2Assigned To => user2
2014-06-24 20:43user2Statusresolved => closed
2015-03-25 08:02user2Relationship addedrelated to 0002279

Notes
(0001193)
user2   
2014-05-30 14:19   
This is crappy.

1) Set clock ahead by a lot (e.g. a month)
2) Run "yum repolist"
3) Set clock back to current time (ntpdate -u time.clearsdn.com)
4) Run "yum repolist"

The yum cache is used despite the large time discrepancy. It would be nice if yum noticed this time travel in invalidated the cache. Create a workaround for now.