[ https://issues.apache.org/jira/browse/IGNITE-6832?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16326205#comment-16326205 ]
Alexey Goncharuk commented on IGNITE-6832: ------------------------------------------ For starters, we need to have a generic method to check the environment and invoke when an unrecoverable exception occurs. > handle IO errors while checkpointing > ------------------------------------ > > Key: IGNITE-6832 > URL: https://issues.apache.org/jira/browse/IGNITE-6832 > Project: Ignite > Issue Type: Bug > Affects Versions: 2.1 > Reporter: Alexander Belyak > Priority: Major > > If we get some IO error (like "No spece left on device") during checkpointing > (GridCacheDatabaseSharedManager$WriteCheckpointPages:2509) node didn't stop > as when get same error while writting WAL log and clients will get some "Long > running cache futures". We must stop node in this case! Better - add some > internal healthcheck and stop node anyway if it won't pass for few times (do > it with different issue). -- This message was sent by Atlassian JIRA (v7.6.3#76005)