- edited description
Server sometimes fails due to possible corrupted cache files
In the directory vt_cache, there are some files named *.vt_cache_pickle They are created upon usage, likely to speed up the turn-around times. However, sometimes the files are corrupted or not readable for the server, and then a "get capabilities" request is answered by "internal server error". This problem is always solved, after the cache files have been deleted.
Should be repaired either by defining an exception or by generally undo the cache mechanism that may not save too much time.
Comments (9)
-
reporter -
first we should identify the reason why that is happening. If it looks like we would need locking then the only option is to skip those cache files.
-
reporter The next time when this happens, we should save the cache files at a different location before deleting. Then ye could better analyse this.
-
We could implement a proper fault handling (i.e. handle the Exception by skipping the cache file) that would allow the server to keep on operating, log the issue, and thus enable both a continued operation and the facility to debug the issue.
-
Next time when it happens, we need the data, And before fixing, we should have a unit test which shows then this error. Don't workaround (try solve) this without a test.
-
- changed milestone to 1.4.0
-
-
- changed status to resolved
refactored wsgi auth handler, fixes
#141,#118→ <<cset d07b7136dd18>>
-
Merged in ReimarBauer/mss (pull request #228)
refactored wsgi auth handler, fixes
#141,#118→ <<cset b0c26af893e5>>
- Log in to comment