[jira] [Comment Edited] (SOLR-5890) Delete silently fails if not sent to shard where document was added
[ https://issues.apache.org/jira/browse/SOLR-5890?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14264553#comment-14264553 ] Ishan Chattopadhyaya edited comment on SOLR-5890 at 1/5/15 12:58 PM: - Removed the broadcast deleteById commands (without __route__ parameter, implicit router) and added a new patch. We can discuss the need for broadcast of deleteById for implicit router (without __route__) in another issue (SOLR-6910). was (Author: ichattopadhyaya): Removed the broadcast deleteById commands (without __route__ parameter, implicit router) and added a new patch. We can discuss the need for broadcast of deleteById for implicit router (without __route__) in another issue. Delete silently fails if not sent to shard where document was added --- Key: SOLR-5890 URL: https://issues.apache.org/jira/browse/SOLR-5890 Project: Solr Issue Type: Bug Components: SolrCloud Affects Versions: 4.7 Environment: Debian 7.4. Reporter: Peter Inglesby Assignee: Noble Paul Labels: difficulty-medium, impact-medium, workaround-exists Fix For: 5.0, Trunk Attachments: 5890_tests.patch, SOLR-5890-without-broadcast.patch, SOLR-5890.patch, SOLR-5890.patch, SOLR-5890.patch, SOLR-5980.patch We have SolrCloud set up with two shards, each with a leader and a replica. We use haproxy to distribute requests between the four nodes. Regardless of which node we send an add request to, following a commit, the newly-added document is returned in a search, as expected. However, we can only delete a document if the delete request is sent to a node in the shard where the document was added. If we send the delete request to a node in the other shard (and then send a commit) the document is not deleted. Such a delete request will get a 200 response, with the following body: {'responseHeader'={'status'=0,'QTime'=7}} Apart from the the very low QTime, this is indistinguishable from a successful delete. -- This message was sent by Atlassian JIRA (v6.3.4#6332) - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] [Comment Edited] (SOLR-5890) Delete silently fails if not sent to shard where document was added
[ https://issues.apache.org/jira/browse/SOLR-5890?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14264553#comment-14264553 ] Ishan Chattopadhyaya edited comment on SOLR-5890 at 1/5/15 12:50 PM: - Removed the broadcast deleteById commands (without __route__ parameter, implicit router) and added a new patch. We can discuss the need for broadcast of deleteById for implicit router (without __route__) in another issue. was (Author: ichattopadhyaya): Removed the broadcast deleteById commands (without __route__parameter, implicit router) and added a new patch. We can discuss the need for broadcast of deleteById for implicit router (without __route__) in another issue. Delete silently fails if not sent to shard where document was added --- Key: SOLR-5890 URL: https://issues.apache.org/jira/browse/SOLR-5890 Project: Solr Issue Type: Bug Components: SolrCloud Affects Versions: 4.7 Environment: Debian 7.4. Reporter: Peter Inglesby Assignee: Noble Paul Labels: difficulty-medium, impact-medium, workaround-exists Fix For: 5.0, Trunk Attachments: 5890_tests.patch, SOLR-5890-without-broadcast.patch, SOLR-5890.patch, SOLR-5890.patch, SOLR-5890.patch, SOLR-5980.patch We have SolrCloud set up with two shards, each with a leader and a replica. We use haproxy to distribute requests between the four nodes. Regardless of which node we send an add request to, following a commit, the newly-added document is returned in a search, as expected. However, we can only delete a document if the delete request is sent to a node in the shard where the document was added. If we send the delete request to a node in the other shard (and then send a commit) the document is not deleted. Such a delete request will get a 200 response, with the following body: {'responseHeader'={'status'=0,'QTime'=7}} Apart from the the very low QTime, this is indistinguishable from a successful delete. -- This message was sent by Atlassian JIRA (v6.3.4#6332) - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] [Comment Edited] (SOLR-5890) Delete silently fails if not sent to shard where document was added
[ https://issues.apache.org/jira/browse/SOLR-5890?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14263581#comment-14263581 ] Ishan Chattopadhyaya edited comment on SOLR-5890 at 1/3/15 6:37 PM: Updated the patch with the above TODO item. This patch now has: 1. deleteById command now has a _route_ parameter. 2. commitWithin wasn't working (at least with the implicit router). Added a fix in SolrCmdDistributor. 3. deleteById command (implicit router) should be distributed to all shard leaders if no _route_ parameter is present. [~noble.paul], [~shalinmangar], please review (review request https://reviews.apache.org/r/29561/). was (Author: ichattopadhyaya): Updated the patch with the above TODO item. This patch now has: 1. deleteById command now has a _route_ parameter. 2. commitWithin wasn't working (at least with the implicit router). Added a fix in SolrCmdDistributor. 3. deleteById command (implicit router) should be distributed to all shard leaders if no _route_ parameter is present. [~noble.paul], [~shalinmangar], please review. Delete silently fails if not sent to shard where document was added --- Key: SOLR-5890 URL: https://issues.apache.org/jira/browse/SOLR-5890 Project: Solr Issue Type: Bug Components: SolrCloud Affects Versions: 4.7 Environment: Debian 7.4. Reporter: Peter Inglesby Assignee: Noble Paul Labels: difficulty-medium, impact-medium, workaround-exists Fix For: 5.0, Trunk Attachments: 5890_tests.patch, SOLR-5890.patch, SOLR-5890.patch, SOLR-5980.patch We have SolrCloud set up with two shards, each with a leader and a replica. We use haproxy to distribute requests between the four nodes. Regardless of which node we send an add request to, following a commit, the newly-added document is returned in a search, as expected. However, we can only delete a document if the delete request is sent to a node in the shard where the document was added. If we send the delete request to a node in the other shard (and then send a commit) the document is not deleted. Such a delete request will get a 200 response, with the following body: {'responseHeader'={'status'=0,'QTime'=7}} Apart from the the very low QTime, this is indistinguishable from a successful delete. -- This message was sent by Atlassian JIRA (v6.3.4#6332) - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] [Comment Edited] (SOLR-5890) Delete silently fails if not sent to shard where document was added
[ https://issues.apache.org/jira/browse/SOLR-5890?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14234208#comment-14234208 ] Ishan Chattopadhyaya edited comment on SOLR-5890 at 12/4/14 1:54 PM: - For supporting _route_ parameter per id in a deleteById request, instead of a _route_ parameter per request, I propose the following changes to the request format. JSON: delete: { id:ID , _route_:route} XML: delete id _route_=shard1123/id id _route_=shard2124/id /delete Similarly, together with UpdateRequest.deleteById(ListString ids), I propose another method: UpdateRequest.deleteById(ListString ids, ListString routes), Here, I don't like the idea of list of encapsulated objects that contains both id and route, hence want to go with two lists. Any suggestions? Also, like UpdateRequest.deleteById(String id), I propose another method: UpdateRequest.deleteById(String id, String route) Can someone please review these suggestions? was (Author: ichattopadhyaya): For supporting _route_ parameter per id in a deleteById request, instead of a _route_ parameter per request, I propose the following changes to the request format. JSON: delete: { id:ID , _route_:route} XML: delete id _route_=shard1123/id id _route_=shard2124/id /delete Similarly, together with UpdateRequest.deleteById(ListString ids), I propose another method: UpdateRequest.deleteById(ListString ids, ListString routes), Here, I don't like the idea of list of encapsulated objects that contains both id and route, hence want to go with two lists. Any suggestions? Also, like UpdateRequest.deleteById(String id), I propose another method: UpdateRequest.deleteById(String id, String route) Delete silently fails if not sent to shard where document was added --- Key: SOLR-5890 URL: https://issues.apache.org/jira/browse/SOLR-5890 Project: Solr Issue Type: Bug Components: SolrCloud Affects Versions: 4.7 Environment: Debian 7.4. Reporter: Peter Inglesby Assignee: Noble Paul Labels: difficulty-medium, impact-medium, workaround-exists Fix For: 5.0, Trunk We have SolrCloud set up with two shards, each with a leader and a replica. We use haproxy to distribute requests between the four nodes. Regardless of which node we send an add request to, following a commit, the newly-added document is returned in a search, as expected. However, we can only delete a document if the delete request is sent to a node in the shard where the document was added. If we send the delete request to a node in the other shard (and then send a commit) the document is not deleted. Such a delete request will get a 200 response, with the following body: {'responseHeader'={'status'=0,'QTime'=7}} Apart from the the very low QTime, this is indistinguishable from a successful delete. -- This message was sent by Atlassian JIRA (v6.3.4#6332) - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] [Comment Edited] (SOLR-5890) Delete silently fails if not sent to shard where document was added
[ https://issues.apache.org/jira/browse/SOLR-5890?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14234211#comment-14234211 ] Ishan Chattopadhyaya edited comment on SOLR-5890 at 12/4/14 1:58 PM: - Whenever it is not possible to identify the shard correctly (either using id or router.field ) the requests should be distributed across the cluster [~noble.paul] For a deleteById (implicit or compositeId router) without a __route__ parameter, do you think it is reasonable to distribute the request across the cluster always? was (Author: ichattopadhyaya): Whenever it is not possible to identify the shard correctly (either using id or router.field ) the requests should be distributed across the cluster For a deleteById (implicit or compositeId router) without a __route__ parameter, do you think it is reasonable to distribute the request across the cluster always? Delete silently fails if not sent to shard where document was added --- Key: SOLR-5890 URL: https://issues.apache.org/jira/browse/SOLR-5890 Project: Solr Issue Type: Bug Components: SolrCloud Affects Versions: 4.7 Environment: Debian 7.4. Reporter: Peter Inglesby Assignee: Noble Paul Labels: difficulty-medium, impact-medium, workaround-exists Fix For: 5.0, Trunk We have SolrCloud set up with two shards, each with a leader and a replica. We use haproxy to distribute requests between the four nodes. Regardless of which node we send an add request to, following a commit, the newly-added document is returned in a search, as expected. However, we can only delete a document if the delete request is sent to a node in the shard where the document was added. If we send the delete request to a node in the other shard (and then send a commit) the document is not deleted. Such a delete request will get a 200 response, with the following body: {'responseHeader'={'status'=0,'QTime'=7}} Apart from the the very low QTime, this is indistinguishable from a successful delete. -- This message was sent by Atlassian JIRA (v6.3.4#6332) - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] [Comment Edited] (SOLR-5890) Delete silently fails if not sent to shard where document was added
[ https://issues.apache.org/jira/browse/SOLR-5890?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14235162#comment-14235162 ] Ishan Chattopadhyaya edited comment on SOLR-5890 at 12/5/14 6:35 AM: - [~shalinmangar] Yes, I tested it (patch attached for the test). Here, as far as I understand, since the router is implicit, there is no hashing of the id performed to find the right shard; instead the deleteById is performed on the shard to which the update request is sent to (unless a specific _route_ param is present). Maybe I'm missing something? was (Author: ichattopadhyaya): [~shalinmangar] Yes, I tested it (patch attached for the test). Here, as far as I understand, since the router is implicit, there is no hashing of the id performed to find the right shard; instead the deleteById is performed on the shard to which the update request is sent to (unless a specific _route_ param is present). Delete silently fails if not sent to shard where document was added --- Key: SOLR-5890 URL: https://issues.apache.org/jira/browse/SOLR-5890 Project: Solr Issue Type: Bug Components: SolrCloud Affects Versions: 4.7 Environment: Debian 7.4. Reporter: Peter Inglesby Assignee: Noble Paul Labels: difficulty-medium, impact-medium, workaround-exists Fix For: 5.0, Trunk Attachments: 5890_tests.patch We have SolrCloud set up with two shards, each with a leader and a replica. We use haproxy to distribute requests between the four nodes. Regardless of which node we send an add request to, following a commit, the newly-added document is returned in a search, as expected. However, we can only delete a document if the delete request is sent to a node in the shard where the document was added. If we send the delete request to a node in the other shard (and then send a commit) the document is not deleted. Such a delete request will get a 200 response, with the following body: {'responseHeader'={'status'=0,'QTime'=7}} Apart from the the very low QTime, this is indistinguishable from a successful delete. -- This message was sent by Atlassian JIRA (v6.3.4#6332) - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] [Comment Edited] (SOLR-5890) Delete silently fails if not sent to shard where document was added
[ https://issues.apache.org/jira/browse/SOLR-5890?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14234208#comment-14234208 ] Ishan Chattopadhyaya edited comment on SOLR-5890 at 12/5/14 6:51 AM: - For supporting _route_ parameter per id in a deleteById request, instead of a _route_ parameter per request, I propose the following changes to the request format. JSON: delete: { id:ID , _route_:route} XML: delete id _route_=shard1123/id id _route_=shard2124/id /delete Also, like UpdateRequest.deleteById(String id), I propose another method: UpdateRequest.deleteById(String id, String route) Can someone please review these suggestions? was (Author: ichattopadhyaya): For supporting _route_ parameter per id in a deleteById request, instead of a _route_ parameter per request, I propose the following changes to the request format. JSON: delete: { id:ID , _route_:route} XML: delete id _route_=shard1123/id id _route_=shard2124/id /delete Similarly, together with UpdateRequest.deleteById(ListString ids), I propose another method: UpdateRequest.deleteById(ListString ids, ListString routes), Here, I don't like the idea of list of encapsulated objects that contains both id and route, hence want to go with two lists. Any suggestions? Also, like UpdateRequest.deleteById(String id), I propose another method: UpdateRequest.deleteById(String id, String route) Can someone please review these suggestions? Delete silently fails if not sent to shard where document was added --- Key: SOLR-5890 URL: https://issues.apache.org/jira/browse/SOLR-5890 Project: Solr Issue Type: Bug Components: SolrCloud Affects Versions: 4.7 Environment: Debian 7.4. Reporter: Peter Inglesby Assignee: Noble Paul Labels: difficulty-medium, impact-medium, workaround-exists Fix For: 5.0, Trunk Attachments: 5890_tests.patch We have SolrCloud set up with two shards, each with a leader and a replica. We use haproxy to distribute requests between the four nodes. Regardless of which node we send an add request to, following a commit, the newly-added document is returned in a search, as expected. However, we can only delete a document if the delete request is sent to a node in the shard where the document was added. If we send the delete request to a node in the other shard (and then send a commit) the document is not deleted. Such a delete request will get a 200 response, with the following body: {'responseHeader'={'status'=0,'QTime'=7}} Apart from the the very low QTime, this is indistinguishable from a successful delete. -- This message was sent by Atlassian JIRA (v6.3.4#6332) - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org