[
https://issues.apache.org/jira/browse/SOLR-14993?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17229875#comment-17229875
]
Allen Sooredoo edited comment on SOLR-14993 at 11/11/20, 9:49 AM:
------------------------------------------------------------------
Hi,
sorry about that, I wanted to do a very quick fix in github and link it here,
but it turns out I need much more time to write all the tests and submit a
proper PR.
Use case:
- I add to every configset a file called id.txt which contains the id of the
configset (positive integer)
- When I want to update it (to add new synonyms for instance), I first use
`zkClient.downloadConfig(configSetName, directory)` (Solrj) to download the
configset and check the id inside the file before proceeding.
Issue:
- If the id fits on 1 byte, the file is created but its content is not
downloaded because of this check:
|private static int copyDataDown(SolrZkClient zkClient, String zkPath, File
file) throws IOException, KeeperException, InterruptedException {
{color:#d73a49}byte{color}[] data = zkClient.getData(zkPath, null, null, true);
if (data != null && data.length > 1) { // There are apparently basically empty
ZNodes.|
https://github.com/apache/lucene-solr/blob/master/solr/solrj/src/java/org/apache/solr/common/cloud/ZkMaintenanceUtils.java
Changing it to `data.length > 0` would solve my problem, however I may need
more time to test that this "empty ZNodes" issue doesn't reappear as a side
effect. Do those "empty ZNodes" have a size of 1 byte? Why not 0?
I'm curious if the check > 1 (instead of the expected >0) is a typo or an
actual fix to a know problem. Do you have more information about that?
was (Author: alsofr):
Hi,
sorry about that, I wanted to do a very quick fix in github and link it here,
but it turns out I need much more time to write all the tests and submit a
proper PR.
Use case:
- I add to every configset a file called id.txt which contains the id of the
configset (positive integer)
- When I want to update it (to add new synonyms for instance), I first use
`zkClient.downloadConfig(configSetName, directory)` (Solrj) to download the
configset and check the id inside the file before proceeding.
Issue:
- If the id fits on 1 byte, the file is created but its content is not
downloaded because of this check:
|private static int copyDataDown(SolrZkClient zkClient, String zkPath, File
file) throws IOException, KeeperException, InterruptedException {
{color:#d73a49}byte{color}[] data = zkClient.getData(zkPath, null, null, true);
if (data != null && data.length > 1) { // There are apparently basically empty
ZNodes.|
Changing it to `data.length > 0` would solve my problem, however I may need
more time to test that this "empty ZNodes" issue doesn't reappear as a side
effect. Do those "empty ZNodes" have a size of 1 byte? Why not 0?
I'm curious if the check > 1 (instead of the expected >0) is a typo or an
actual fix to a know problem. Do you have more information about that?
> Unable to download zookeeper files of 1byte in size
> ---------------------------------------------------
>
> Key: SOLR-14993
> URL: https://issues.apache.org/jira/browse/SOLR-14993
> Project: Solr
> Issue Type: Bug
> Security Level: Public(Default Security Level. Issues are Public)
> Components: SolrCloud, SolrJ
> Affects Versions: 8.5.1
> Reporter: Allen Sooredoo
> Priority: Minor
>
> When downloading a file from Zookeeper using the Solrj client, files of size
> 1 byte are ignored.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org
For additional commands, e-mail: issues-h...@lucene.apache.org
