So I have on azure umbraco 7.5.13 site. Its a master slave setup. So we have backoffice as master and frontend as slave. Front end does not have /umbraco access. I have followed instructions on our and have set up master slave registrations.
Two things when my site loads I do not see any entries in the umbracoserver table. The other thing is the internal index will not build. External one builds fine. The error log when rebuilding internal index does give the following error:
at Examine.LuceneEngine.Providers.LuceneIndexer.RebuildIndex() in X:\Projects\Examine\Examine\src\Examine\LuceneEngine\Providers\LuceneIndexer.cs:line 752
at UmbracoExamine.UmbracoContentIndexer.RebuildIndex()
at Umbraco.Web.WebServices.ExamineManagementApiController.PostRebuildIndex(String indexerName)
ClientConnectionId:ed65fabc-42be-4c41-9096-5dfee842d32c
Error Number:-2,State:0,Class:11
ClientConnectionId before routing:88382140-7485-4377-9d3a-ab58663fea6e
Routing Destination:xxxxxxxx,11023
I have stripped out the db server and replaced with xxxxxxx. On the backoffice master I can login and get to content tree etc so its hitting the db with connection string I have in the web.config. Anyone seen this before?
Ah yes, S0 is tricky, it does work as long as you increase the timeout but S1 is recommended minimum for Umbraco. If you can try to get a elastic pool with a few different projects in it, can make it cheaper and you have more burstable eDTUs.
Some people have asked that we continue populating this table regardless for informational purposes and it's something I'm sure we can do but at the moment that isn't a priority (PRs welcome! :)
Have you tried out 7.6.5 ? IIRC there are performance enhancements in versions greater than the one you are using.
Here is our ExamineSettings.config Azure transform file:
<?xml version="1.0"?>
<!-- For more information on using app.config transformation visit http://go.microsoft.com/fwlink/?LinkId=125889 -->
<Examine xmlns:xdt="http://schemas.microsoft.com/XML-Document-Transform">
<ExamineIndexProviders>
<providers>
<add directoryFactory="Examine.LuceneEngine.Directories.TempEnvDirectoryFactory,Examine" xdt:Transform="SetAttributes" />
</providers>
</ExamineIndexProviders>
</Examine>
Load balance on azure index rebuild issue
So I have on azure umbraco 7.5.13 site. Its a master slave setup. So we have backoffice as master and frontend as slave. Front end does not have /umbraco access. I have followed instructions on our and have set up master slave registrations.
Two things when my site loads I do not see any entries in the umbracoserver table. The other thing is the internal index will not build. External one builds fine. The error log when rebuilding internal index does give the following error:
I have stripped out the db server and replaced with xxxxxxx. On the backoffice master I can login and get to content tree etc so its hitting the db with connection string I have in the web.config. Anyone seen this before?
Regards
Ismail
So a bit more information it looks like http://issues.umbraco.org/issue/U4-7869 we do have 70k items altogether.
I have also found that we can increase time out https://github.com/umbraco/Umbraco-CMS/pull/1097/files will give that a go see if that works.
Give that a try, also what type of DB (S0, S1, etc) do you have and have you checked to see how the DTU utilisation is when building the indexes?
Jeavon,
I am slowly bumping up dtus its making a difference also increased timeout but think i just need to keep going till it builds. cheers.
was on s0 im on s2 now with 50 dtus
Ismail
Jeavon,
That did the trick s2 50 dtus.
Regards
Ismail
Ah yes, S0 is tricky, it does work as long as you increase the timeout but S1 is recommended minimum for Umbraco. If you can try to get a elastic pool with a few different projects in it, can make it cheaper and you have more burstable eDTUs.
umbracoserver is only in use for master election, if you follow the recommended practice here https://our.umbraco.org/documentation/Getting-Started/Setup/Server-Setup/load-balancing/flexible#scheduling-and-master-election then master election does not occur and therefore this table is not populated.
Some people have asked that we continue populating this table regardless for informational purposes and it's something I'm sure we can do but at the moment that isn't a priority (PRs welcome! :)
Have you tried out 7.6.5 ? IIRC there are performance enhancements in versions greater than the one you are using.
In the flexible load balancing docs there is setting:
do we need this? We have got the azure bits setup. So atm our config looks like:
regards
Ismail
Shannon,
Not tried updating as client wont pay for it so we are stuck on 7.5. Increasing timeouts and dtus has got the indexes rebuilding.
Regards
Ismail
If you have Examine v0.1.80+ then use the directoryFactory otherwise what you have is what is recommended for pre v0.1.80
i got 0.182 so keep both azure bits and directoryfactory or just directoryfactory?
Regards
Ismail
Just directoryFactory
Here is our ExamineSettings.config Azure transform file:
Make everything so much simpler than it used to be :)
SyncTempEnvDirectoryFactory is available in Examine v0.1.80+
This is equivalent to what you had before but "better"
TempEnvDirectoryFactory is only available in Examine v0.1.83+
This is even better as it doesn't need to sync and you don't have to tokenise the index paths
is working on a reply...