Re: knapsack use case
…just a small update in case anyone was wondering: I completed an elasticsearch-knapsack export to file from a 0.90.7 (with the plugin built with 0.90.7 dependencies) that was correctly re-imported in a 1.4.2 instance with the latest plugin version, including settings and aliases. I just checked out the source from github and changed 2-3 lines due to API changes and assembled the plugin to be installed on a 0.90.7 instance. Just wanted to say thanks! Best, Matteo On Thu, Jul 31, 2014 at 11:33 AM, joergpra...@gmail.com < joergpra...@gmail.com> wrote: > Snapshot/restore is always recommended, but is a 1.0 feature. This is a > standard API of ES and well supported by the ES team. With that, you can > handle all kinds of index data safely on a binary level, fully and > incrementally. > > Knapsack plugin is for document export/import only. I wrote it to > transport _source data harvested over a long time period from a < 1.0 > system to a production system. It works on _source or stored fields only. > It uses search/query and bulk indexing API without snapshots, so it is up > to the admin to stop index writes while knapsack runs. There is also a > lookup of index settings and mappings, this information is also included in > the export archive file, and re-applied at import time. But, there is no > check if these settings/mappings can be applied on the target successfully, > this is left to the admin to prepare plugins, analyzers, etc. Aliases are > not transported but this is a good idea for improvement. > > Currently, knapsack plugin does not work on ES 1.3 but I am progressing to > implement this. I am adding a Java-level API. Currently it is REST only. > > Jörg > > > On Thu, Jul 31, 2014 at 11:05 AM, Matteo Moci wrote: > >> Hi All, >> I have some questions about the knapsack plugin [1]. >> >> My idea to use the tool to do a backup to a file, starting from a 0.90.x >> instance and then restore it on a different 1.2.x or 1.3.x instance. I see >> it can't be done directly, copying to a local/remote cluster. >> >> Would it work doing an intermediate step with a file? >> Or the backup still has metadata about the es version it was generated >> from, making it impossible? >> >> Is the snapshot and restore feature [2] useful in my use case, or not? >> >> Is the knapsack plugin able to backup and restore also aliases and >> mappings, or do I have to manually migrate them before restoring data? >> >> Thanks for the patience and the great work! >> Matteo >> >> [1] https://github.com/jprante/elasticsearch-knapsack >> [2] >> http://www.elasticsearch.org/guide/en/elasticsearch/reference/current/modules-snapshots.html >> >> -- >> Matteo Moci >> http://mox.fm >> >> -- >> You received this message because you are subscribed to the Google Groups >> "elasticsearch" group. >> To unsubscribe from this group and stop receiving emails from it, send an >> email to elasticsearch+unsubscr...@googlegroups.com. >> To view this discussion on the web visit >> https://groups.google.com/d/msgid/elasticsearch/CAONgFZ60jWViqzRVO6_U-rYo6dUzunE3ojv%2BR5U8HX1Lwp4PdA%40mail.gmail.com >> <https://groups.google.com/d/msgid/elasticsearch/CAONgFZ60jWViqzRVO6_U-rYo6dUzunE3ojv%2BR5U8HX1Lwp4PdA%40mail.gmail.com?utm_medium=email&utm_source=footer> >> . >> For more options, visit https://groups.google.com/d/optout. >> > > -- > You received this message because you are subscribed to the Google Groups > "elasticsearch" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to elasticsearch+unsubscr...@googlegroups.com. > To view this discussion on the web visit > https://groups.google.com/d/msgid/elasticsearch/CAKdsXoGrXthJjCchEf2oyvXKnSZyBp31nvnAeXwAZJaEkvnT5Q%40mail.gmail.com > <https://groups.google.com/d/msgid/elasticsearch/CAKdsXoGrXthJjCchEf2oyvXKnSZyBp31nvnAeXwAZJaEkvnT5Q%40mail.gmail.com?utm_medium=email&utm_source=footer> > . > For more options, visit https://groups.google.com/d/optout. > -- Matteo Moci http://mox.fm -- You received this message because you are subscribed to the Google Groups "elasticsearch" group. To unsubscribe from this group and stop receiving emails from it, send an email to elasticsearch+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/elasticsearch/CAONgFZ6PvJyeF04ERyeb26LhXoHR%3DMMs5sc5KV2ASLF_UK6b%3Dw%40mail.gmail.com. For more options, visit https://groups.google.com/d/optout.
knapsack use case
Hi All, I have some questions about the knapsack plugin [1]. My idea to use the tool to do a backup to a file, starting from a 0.90.x instance and then restore it on a different 1.2.x or 1.3.x instance. I see it can't be done directly, copying to a local/remote cluster. Would it work doing an intermediate step with a file? Or the backup still has metadata about the es version it was generated from, making it impossible? Is the snapshot and restore feature [2] useful in my use case, or not? Is the knapsack plugin able to backup and restore also aliases and mappings, or do I have to manually migrate them before restoring data? Thanks for the patience and the great work! Matteo [1] https://github.com/jprante/elasticsearch-knapsack [2] http://www.elasticsearch.org/guide/en/elasticsearch/reference/current/modules-snapshots.html -- Matteo Moci http://mox.fm -- You received this message because you are subscribed to the Google Groups "elasticsearch" group. To unsubscribe from this group and stop receiving emails from it, send an email to elasticsearch+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/elasticsearch/CAONgFZ60jWViqzRVO6_U-rYo6dUzunE3ojv%2BR5U8HX1Lwp4PdA%40mail.gmail.com. For more options, visit https://groups.google.com/d/optout.
Re: SearchContextMissingException when doing scan search [0.90.7-0.90.13]
Well spotted Joerg! I knew the scroll id was changing, anyway in the end I coded it wrong. Thanks and keep up the good work! Matteo On Mon, Apr 28, 2014 at 1:29 PM, joergpra...@gmail.com < joergpra...@gmail.com> wrote: > Reviewing your code, you have made a small mistake. > > In the while loop, do not use currentScanSearchResp.getScrollId() > > > https://gist.github.com/mox601/545c7176785ef209f7f3#file-scroll-search-java-L59 > > This can get outdated in the while loop. > > Instead, use searchResponse.getScrollId() > > Background info: the scroll ID changes from response to response! > > Jörg > > > > On Mon, Apr 28, 2014 at 12:44 PM, Matteo Moci wrote: > >> Hi Jorg, >> Investigating on what call the exception relates to, >> I ran the search on 2 types, and I'll show you the merged logs of my >> client and the one printed on the node: >> >> [2014-04-28 12:09:07,*438*][DEBUG][action.search.type ] >> [moxbook-pro] [2] Failed to execute query phase >> 2014-04-28 12:09:07,*442*/CEST [main] INFO finished scrolling, hits >> length <=0 >> [2014-04-28 12:09:07,*455*][DEBUG][action.search.type ] >> [moxbook-pro] [3] Failed to execute query phase >> 2014-04-28 12:09:07,*456*/CEST [main] INFO finished scrolling, hits >> length <=0 >> >> The client logs are printed by the client code (running on the same >> machine), >> once per type, after exiting the while (hits.length > 0) scroll's cycle >> and >> before calling the clear scroll at the end. >> >> So to wrap up, according to what I see, the error message >> of the failing execute query phase happens before the clear scroll. >> >> Following another path and reading your first email, >> I added to the configuration file elasticsearch.yml the line: >> >> search.keep_alive_interval: 1H >> hoping to set the search keep alive explicitly on the node. >> >> Starting the node like this: >> $ ./elasticsearch -f -Des.logger.level=DEBUG >> gives this log [1], but can't find mention of the keep alive interval I >> set up. >> >> Is it the right parameter name, and should it help? >> >> Thanks, >> Matteo >> >> [1] https://gist.github.com/mox601/11368114 >> >> >> >> On Mon, Apr 28, 2014 at 11:53 AM, joergpra...@gmail.com < >> joergpra...@gmail.com> wrote: >> >>> OK, the "new Scroll(...)" should'nt make any difference >>> >>> Maybe the cause for theerror message is the clear scroll call at the >>> end? If so, it shouldn't be serious. >>> >>> Jörg >>> >>> -- >>> You received this message because you are subscribed to the Google >>> Groups "elasticsearch" group. >>> To unsubscribe from this group and stop receiving emails from it, send >>> an email to elasticsearch+unsubscr...@googlegroups.com. >>> To view this discussion on the web visit >>> https://groups.google.com/d/msgid/elasticsearch/CAKdsXoGUec-BG2%2BJz9VOjXMjjwEeqxqMyA2XNusWCmMBKt0f8w%40mail.gmail.com<https://groups.google.com/d/msgid/elasticsearch/CAKdsXoGUec-BG2%2BJz9VOjXMjjwEeqxqMyA2XNusWCmMBKt0f8w%40mail.gmail.com?utm_medium=email&utm_source=footer> >>> . >>> >>> For more options, visit https://groups.google.com/d/optout. >>> >> >> >> >> -- >> Matteo Moci >> http://mox.fm >> >> -- >> You received this message because you are subscribed to the Google Groups >> "elasticsearch" group. >> To unsubscribe from this group and stop receiving emails from it, send an >> email to elasticsearch+unsubscr...@googlegroups.com. >> To view this discussion on the web visit >> https://groups.google.com/d/msgid/elasticsearch/CAONgFZ5gcpx2nXUZ31qHWrLb_LB8vpXPCJsmUH9-HAMF34HDmw%40mail.gmail.com<https://groups.google.com/d/msgid/elasticsearch/CAONgFZ5gcpx2nXUZ31qHWrLb_LB8vpXPCJsmUH9-HAMF34HDmw%40mail.gmail.com?utm_medium=email&utm_source=footer> >> . >> >> For more options, visit https://groups.google.com/d/optout. >> > > -- > You received this message because you are subscribed to the Google Groups > "elasticsearch" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to elasticsearch+unsubscr...@googlegroups.com. > To view this discussion on the web visit > https://groups.google.com/d/msgid/elasticsearch/CAKdsXoH1pfcM3UgBmNe9Yq%2BS9S%3DOY7eo_P1urGs5Kj2wORv%2BWQ%40mail.gmail.com<https://groups.google.com/d/msgid/elasticsearch/CAKdsXoH1pfcM3UgBmNe9Yq%2BS9S%3DOY7eo_P1urGs5Kj2wORv%2BWQ%40mail.gmail.com?utm_medium=email&utm_source=footer> > . > > For more options, visit https://groups.google.com/d/optout. > -- Matteo Moci http://mox.fm -- You received this message because you are subscribed to the Google Groups "elasticsearch" group. To unsubscribe from this group and stop receiving emails from it, send an email to elasticsearch+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/elasticsearch/CAONgFZ6XDw4c_2Lpax%3DBcM%2BTvtNCP4yfssbOTHs8R7_p1wiXmQ%40mail.gmail.com. For more options, visit https://groups.google.com/d/optout.
Re: SearchContextMissingException when doing scan search [0.90.7-0.90.13]
Hi Jorg, Investigating on what call the exception relates to, I ran the search on 2 types, and I'll show you the merged logs of my client and the one printed on the node: [2014-04-28 12:09:07,*438*][DEBUG][action.search.type ] [moxbook-pro] [2] Failed to execute query phase 2014-04-28 12:09:07,*442*/CEST [main] INFO finished scrolling, hits length <=0 [2014-04-28 12:09:07,*455*][DEBUG][action.search.type ] [moxbook-pro] [3] Failed to execute query phase 2014-04-28 12:09:07,*456*/CEST [main] INFO finished scrolling, hits length <=0 The client logs are printed by the client code (running on the same machine), once per type, after exiting the while (hits.length > 0) scroll's cycle and before calling the clear scroll at the end. So to wrap up, according to what I see, the error message of the failing execute query phase happens before the clear scroll. Following another path and reading your first email, I added to the configuration file elasticsearch.yml the line: search.keep_alive_interval: 1H hoping to set the search keep alive explicitly on the node. Starting the node like this: $ ./elasticsearch -f -Des.logger.level=DEBUG gives this log [1], but can't find mention of the keep alive interval I set up. Is it the right parameter name, and should it help? Thanks, Matteo [1] https://gist.github.com/mox601/11368114 On Mon, Apr 28, 2014 at 11:53 AM, joergpra...@gmail.com < joergpra...@gmail.com> wrote: > OK, the "new Scroll(...)" should'nt make any difference > > Maybe the cause for theerror message is the clear scroll call at the end? > If so, it shouldn't be serious. > > Jörg > > -- > You received this message because you are subscribed to the Google Groups > "elasticsearch" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to elasticsearch+unsubscr...@googlegroups.com. > To view this discussion on the web visit > https://groups.google.com/d/msgid/elasticsearch/CAKdsXoGUec-BG2%2BJz9VOjXMjjwEeqxqMyA2XNusWCmMBKt0f8w%40mail.gmail.com<https://groups.google.com/d/msgid/elasticsearch/CAKdsXoGUec-BG2%2BJz9VOjXMjjwEeqxqMyA2XNusWCmMBKt0f8w%40mail.gmail.com?utm_medium=email&utm_source=footer> > . > > For more options, visit https://groups.google.com/d/optout. > -- Matteo Moci http://mox.fm -- You received this message because you are subscribed to the Google Groups "elasticsearch" group. To unsubscribe from this group and stop receiving emails from it, send an email to elasticsearch+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/elasticsearch/CAONgFZ5gcpx2nXUZ31qHWrLb_LB8vpXPCJsmUH9-HAMF34HDmw%40mail.gmail.com. For more options, visit https://groups.google.com/d/optout.
Re: SearchContextMissingException when doing scan search [0.90.7-0.90.13]
Hi Jorg, Thanks for the reply. If that helps, I linked a snippet with the code at [1]. In addition, the test I did was on 1 index, 2 shards with 0 replicas, see the status at [2] and settings at [3]. Ho can I provide you more details? Thanks, Matteo [1] https://gist.github.com/mox601/545c7176785ef209f7f3 [2] https://gist.github.com/mox601/fd806317dbe1d89ff12c [3] https://gist.github.com/mox601/11366726 On Mon, Apr 28, 2014 at 10:56 AM, joergpra...@gmail.com < joergpra...@gmail.com> wrote: > The message is serious because it signals that your scan/scroll search has > prematurely ended and responses may be dropped (or not). > > It would be nice to show us your code and your shard settings. > > search.keep_alive_interval is the parameter that can influence the search > reaper, not search operation threading. The reaper runs each minute and > might run too frequently. While not sure about your setting of scan/scroll > lifetime, I can't tell more. > > Jörg > > > -- > You received this message because you are subscribed to the Google Groups > "elasticsearch" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to elasticsearch+unsubscr...@googlegroups.com. > To view this discussion on the web visit > https://groups.google.com/d/msgid/elasticsearch/CAKdsXoHfGVg3xAeFXASg9%3DuiWr8w0pH24b6ucDvAfQeY_JC%3DPQ%40mail.gmail.com<https://groups.google.com/d/msgid/elasticsearch/CAKdsXoHfGVg3xAeFXASg9%3DuiWr8w0pH24b6ucDvAfQeY_JC%3DPQ%40mail.gmail.com?utm_medium=email&utm_source=footer> > . > For more options, visit https://groups.google.com/d/optout. > -- Matteo Moci http://mox.fm -- You received this message because you are subscribed to the Google Groups "elasticsearch" group. To unsubscribe from this group and stop receiving emails from it, send an email to elasticsearch+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/elasticsearch/CAONgFZ7cMtUW%3Dyco3EMSA_2p9NPFN3KRAWuiaCA6MGHWv5R0kA%40mail.gmail.com. For more options, visit https://groups.google.com/d/optout.
SearchContextMissingException when doing scan search [0.90.7-0.90.13]
Hi All, First of all, thanks for the nice work! I found out that during a simple scan search (10 docs of 1 type, on 1 node, 1 index with 2 shards, no replicas, both v0.90.7 and v0.90.13), I _always_ receive this exception in the node logs (and it happens once per type when I have more): [2014-04-28 10:22:07,937][DEBUG][action.search.type ] [moxbook] [45] Failed to execute query phase org.elasticsearch.search.SearchContextMissingException: No search context found for id [45] at org.elasticsearch.search.SearchService.findContext(SearchService.java:460) at org.elasticsearch.search.SearchService.executeScan(SearchService.java:211) at org.elasticsearch.search.action.SearchServiceTransportAction.sendExecuteScan(SearchServiceTransportAction.java:474) at org.elasticsearch.action.search.type.TransportSearchScrollScanAction$AsyncAction.executePhase(TransportSearchScrollScanAction.java:210) at org.elasticsearch.action.search.type.TransportSearchScrollScanAction$AsyncAction$2.run(TransportSearchScrollScanAction.java:180) at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1145) at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:615) at java.lang.Thread.run(Thread.java:722) What worries me is that apparently in my tests, it is able to scan all the docs with no missing records, so I don't know if it's "normal" to have logs like this or if in some cases it will lead to problems. I already saw that there's a closed issue at [5165], but I don't know if the fix was already included in the releases I tried, 0.90.7 (surely not) and 0.90.13 (?). Changing SearchOperationThreading doesn't help. You can see the relevant snippet I used to get the exception at [1], maybe there is something missing there? Could someone help to find out if it's ok, or where the fix was introduced? PS: I didn't upgrade to 1.x yet. Best, Matteo [5165] https://github.com/elasticsearch/elasticsearch/issues/5165 [1] https://gist.github.com/mox601/545c7176785ef209f7f3 -- Matteo Moci http://mox.fm -- You received this message because you are subscribed to the Google Groups "elasticsearch" group. To unsubscribe from this group and stop receiving emails from it, send an email to elasticsearch+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/elasticsearch/CAONgFZ5_Mx-3be461Pu2Sgn4jp0N4f_gbjKAGh2nwUf9vRj5pw%40mail.gmail.com. For more options, visit https://groups.google.com/d/optout.