cloudstack-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From jburwell <...@git.apache.org>
Subject [GitHub] cloudstack issue #1762: CLOUDSTACK-9595 Transactions are not getting retried...
Date Mon, 14 Nov 2016 22:10:55 GMT
Github user jburwell commented on the issue:

    https://github.com/apache/cloudstack/pull/1762
  
    @serg38 that is not a safe assumption.  Transactions often span multiple statements and
methods across DAOs.  `TransactionLegacy` has a transaction stacking/nested model that further
occludes when a transaction actually completely.
    
    Deadlocks are a severe problem that need to be fixed.  Unfortunately, this patch would
do more harm than good as it would eventually corrupt the database.   In, and of themselves,
retries are also a very expensive solution to the problem both in terms of the engineering
effort required to do it properly and the extra stress placed on the database to perform additional
work that will likely fail.  Furthermore, a generic **and** correct retry mechanism is a very
difficult thing to write.  Given the way transaction boundaries are managed in ACS, I think
such an effort would be nearly impossible.
    
    In a properly written application, deadlocks should very rarely, if ever, occur.  Their
presence is a symptom of improper transaction handling and/or poor lock management problems.
  Therefore, my suggestion is that we change this patch to log details about the context in
which deadlocks occur.  We can then use this information to identify the areas in ACS where
these contention problems are location and fix the root cause.


---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastructure@apache.org or file a JIRA ticket
with INFRA.
---

Mime
View raw message