You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@sentry.apache.org by "Alexander Kolbasov (JIRA)" <ji...@apache.org> on 2017/06/15 04:52:00 UTC
[jira] [Assigned] (SENTRY-1804) Sentry server can be more efficient
in handling full snapshot from HMS
[ https://issues.apache.org/jira/browse/SENTRY-1804?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Alexander Kolbasov reassigned SENTRY-1804:
------------------------------------------
Assignee: Alexander Kolbasov
> Sentry server can be more efficient in handling full snapshot from HMS
> ----------------------------------------------------------------------
>
> Key: SENTRY-1804
> URL: https://issues.apache.org/jira/browse/SENTRY-1804
> Project: Sentry
> Issue Type: Sub-task
> Components: Sentry
> Affects Versions: sentry-ha-redesign
> Reporter: Alexander Kolbasov
> Assignee: Alexander Kolbasov
> Labels: sentry-ha
>
> When Sentry server creates a full snapshot, it does the following:
> 1) Calls {{fetchFullUpdate()}} to get a full snapshot. This returns {{Map<String, Set<String>>}}
> 2) Goes through all entries and converts them to a different format, allocating memory again for each element:
> {code}
> for (Map.Entry<String, Set<String>> authzPath : authzPaths.entrySet()) {
> createAuthzPathsMappingCore(pm, nextHmsImageID, authzPath.getKey(), authzPath.getValue());
> }
> {code}
> The original full snapshot is then garbage collected and the new one persisted.
> Instead, {{fetchFullUpdate()}} should just create the full snapshot immediately in the final form. This way we do not have to allocate twice as much memory as needed.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)