I just retried with 3.1.17 and was able to reproduce the issue you reported.
I then tried the same with 3.1.18 and it worked there.
So it should be fixed with the 3.1.18 release.
Am Donnerstag, 20. April 2017 09:12:32 UTC+2 schrieb Jan:
> Hi Scott,
> it will hopefully turn out to be a side-effect of issue #2422 (
> which will be fixed in 3.1.18 (the next release).
> It would be very helpful if you could try that release when it comes out.
> I expect it to be released this week.
> Best regards
> Am Mittwoch, 19. April 2017 18:31:01 UTC+2 schrieb Scott B.:
>> Thanks for the reply. I have the issue on three different ArangoDB
>> servers, all running on Ubuntu 16.04 LTS.
>> If you don't see the issue, then I'm beginning to suspect it is caused by
>> some weird edge case, perhaps in our data. I've sent you a few attachments
>> and a more detailed explanation that might help narrow down the issue.
>> This isn't a critical issue for us (hash indexes work fine for now), just
>> very strange.
>> On Tuesday, April 18, 2017 at 5:53:53 AM UTC-6, Jan wrote:
>>> Hi Scott,
>>> I just tried creating a persistent index via the web UI. The index was
>>> on field "arrayName[*].propertyName" as suggested by you.
>>> The index was actually created and it seems to work for me. The type of
>>> the index is also "persistent" when I look at the index in detail later.
>>> I am using 3.1.17 with the latest changes from the 3.1 branch that have
>>> not been released yet.
>>> As it seems to work there, I suspect the issue you are seeing is caused
>>> by the same issue as reported here:
>>> Note that this issue has been fixed in 3.1.18, which will be released
>>> Best regards
>>> Am Donnerstag, 13. April 2017 23:36:29 UTC+2 schrieb Scott B.:
>>>> Using the ArangoDB web UI, I used to be able to create persistent
>>>> indexes on the properties of arrays by using: arrayName[*].propertyName
>>>> However, now (in 3.1.17) that no longer appears to actually create a
>>>> persistent index. It has been a while since I've modified indexes, so I'm
>>>> not sure which version I last used when it did work. I am not marking
>>>> as unique or sparse.
>>>> However, I can still use the same notation to create a hash
>>>> (non-persistent) index. Did something change with regard to persistent
>>>> indexes I'm unaware of, or is this a bug?
You received this message because you are subscribed to the Google Groups
To unsubscribe from this group and stop receiving emails from it, send an email
For more options, visit https://groups.google.com/d/optout.