[jira] [Updated] (IGNITE-20207) Improve the error handling
[ https://issues.apache.org/jira/browse/IGNITE-20207?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ivan Gagarkin updated IGNITE-20207: --- Description: The current implementation of org.apache.ignite.internal.network.file.FileTransferService doesn't provide recovery functionality. In case of any error the transfer should be started from scratch. We need to define cases when Ignite can provide recovery and implement this functionality. (was: The current implementation of org.apache.ignite.internal.network.file.FileTransferService doesn't provide recovery functionality. Any error during file transfer leads to repeating the transfer from scratch. We need to define cases when Ignite can provide recovery and implement this functionality. ) > Improve the error handling > -- > > Key: IGNITE-20207 > URL: https://issues.apache.org/jira/browse/IGNITE-20207 > Project: Ignite > Issue Type: Improvement >Reporter: Ivan Gagarkin >Priority: Major > Labels: ignite-3 > > The current implementation of > org.apache.ignite.internal.network.file.FileTransferService doesn't provide > recovery functionality. In case of any error the transfer should be started > from scratch. We need to define cases when Ignite can provide recovery and > implement this functionality. -- This message was sent by Atlassian Jira (v8.20.10#820010)
[jira] [Updated] (IGNITE-20207) Improve the error handling
[ https://issues.apache.org/jira/browse/IGNITE-20207?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ivan Gagarkin updated IGNITE-20207: --- Description: The current implementation of org.apache.ignite.internal.network.file.FileTransferService doesn't provide recovery functionality. Any error during file transfer leads to repeating the transfer from scratch. We need to define cases when Ignite can provide recovery and implement this functionality. (was: The current implementation of org.apache.ignite.internal.network.file.FileTransferService doesn't provide recovery functionality. Any error during file transfer leads to repeating the transfer from scratch. ) > Improve the error handling > -- > > Key: IGNITE-20207 > URL: https://issues.apache.org/jira/browse/IGNITE-20207 > Project: Ignite > Issue Type: Improvement >Reporter: Ivan Gagarkin >Priority: Major > Labels: ignite-3 > > The current implementation of > org.apache.ignite.internal.network.file.FileTransferService doesn't provide > recovery functionality. Any error during file transfer leads to repeating the > transfer from scratch. We need to define cases when Ignite can provide > recovery and implement this functionality. -- This message was sent by Atlassian Jira (v8.20.10#820010)
[jira] [Updated] (IGNITE-20207) Improve the error handling
[
https://issues.apache.org/jira/browse/IGNITE-20207?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Ivan Gagarkin updated IGNITE-20207:
---
Description: The current implementation of
org.apache.ignite.internal.network.file.FileTransferService doesn't provide
recovery functionality. Any error during file transfer leads to repeat the
transfer from scratch. (was: The current implementation of
{{org.apache.ignite.internal.network.file.ChunkedFileWriter}} compares the file
pointer with the offset of the received file chunk. If they are equal, the
chunk is written to the disk; if not, the chunk is placed in the queue, and it
will be written when all previous chunks have been written.
It might be more efficient to write chunks instantly.
We should investigate this approach and improve the implementation.)
> Improve the error handling
> --
>
> Key: IGNITE-20207
> URL: https://issues.apache.org/jira/browse/IGNITE-20207
> Project: Ignite
> Issue Type: Improvement
>Reporter: Ivan Gagarkin
>Priority: Major
> Labels: ignite-3
>
> The current implementation of
> org.apache.ignite.internal.network.file.FileTransferService doesn't provide
> recovery functionality. Any error during file transfer leads to repeat the
> transfer from scratch.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
[jira] [Updated] (IGNITE-20207) Improve the error handling
[ https://issues.apache.org/jira/browse/IGNITE-20207?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ivan Gagarkin updated IGNITE-20207: --- Description: The current implementation of org.apache.ignite.internal.network.file.FileTransferService doesn't provide recovery functionality. Any error during file transfer leads to repeating the transfer from scratch. (was: The current implementation of org.apache.ignite.internal.network.file.FileTransferService doesn't provide recovery functionality. Any error during file transfer leads to repeat the transfer from scratch. ) > Improve the error handling > -- > > Key: IGNITE-20207 > URL: https://issues.apache.org/jira/browse/IGNITE-20207 > Project: Ignite > Issue Type: Improvement >Reporter: Ivan Gagarkin >Priority: Major > Labels: ignite-3 > > The current implementation of > org.apache.ignite.internal.network.file.FileTransferService doesn't provide > recovery functionality. Any error during file transfer leads to repeating the > transfer from scratch. -- This message was sent by Atlassian Jira (v8.20.10#820010)
[jira] [Updated] (IGNITE-20207) Improve the error handling
[
https://issues.apache.org/jira/browse/IGNITE-20207?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Ivan Gagarkin updated IGNITE-20207:
---
Summary: Improve the error handling (was: Improve the writing of files in
FileTransferService)
> Improve the error handling
> --
>
> Key: IGNITE-20207
> URL: https://issues.apache.org/jira/browse/IGNITE-20207
> Project: Ignite
> Issue Type: Improvement
>Reporter: Ivan Gagarkin
>Priority: Major
> Labels: ignite-3
>
> The current implementation of
> {{org.apache.ignite.internal.network.file.ChunkedFileWriter}} compares the
> file pointer with the offset of the received file chunk. If they are equal,
> the chunk is written to the disk; if not, the chunk is placed in the queue,
> and it will be written when all previous chunks have been written.
> It might be more efficient to write chunks instantly.
> We should investigate this approach and improve the implementation.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
