Errors when answering by email

Meta Elastic
1

I'm getting this now:

We're sorry, but your email message to ["discuss+35309678d2f68e7f4eeba30249bfc97c@elastic.co"] (titled Re: [Elasticsearch] Failed to send ping to) didn't work.

The provided reply key is invalid or unknown, so we don't know what this email is in reply to. Contact a staff member.

3

This morning I relied to messages in two topics, but my reply never was visible visible in the forum. I didn't get an error like dadoonet is reporting.

Threads:

5

Got again an error this morning (100% of my answers get this response):

We're sorry, but your email message to ["discuss+6b3819b9b4cdfba09740f57fcfc7a6e7@elastic.co"] (titled Re: [Elastic In Your Native Tongue/Discussions en français] Combiner deux résultats d'aggregations et les diviser) didn't work.

The provided reply key is invalid or unknown, so we don't know what this email is in reply to. Contact a staff member.

6

Me too!

From: noreply@discuss.elastic.co
Date: Tue, Jun 2, 2015 at 9:05 AM
Subject: [Elastic Discuss] Email issue -- Unknown Reply Key
To: nik9000@gmail.com

We're sorry, but your email message to ["discuss+770f92b92818289cfdfc164940a8faae@elastic.co"] (titled Re: [Elasticsearch] Are segments stores on heap or off heap?) didn't work.

The provided reply key is invalid or unknown, so we don't know what this email is in reply to. Contact a staff member.

7

I've escalated this directly to the team at Discourse who will help us with
troubleshooting. More news as we have it.

Updated this thread:
Interestingly enough, I was able to reply via email with no issues. I am posting this update directly via the web interface.

9

I get the email issues inconsistently. Maybe half or 1/4 of the time.

11

Cool, will do. Thanks.

12

Same here - below is the content of the email (and the key) in case it helps:

We're sorry, but your email message to ["discuss+083d324519b9e473a07d299150d3dd2b@elastic.co"]
(titled Re: [Hadoop and Elasticsearch] Insert data from hive to elasticsearch) didn't work.

The provided reply key is invalid or unknown, so we don't know what this email is in reply to. Contact a staff member.

If the key is there, sounds like a visibility/concurrency/race-condition.

13

I hit this again this morning. Its a nasty problem! I don't like losing half my posts. As much as the forum has been a success these failures are demoralizing and make me wish I had google groups again.

Maybe do a backoff/retry thing? Add tons of logging? All the normal stuff I guess.

15

Just to help debug this, if you reply a second time with the exact same reply does it go through?

16

It just occurred for me and no, resending the email does not work.

This is the thread I'm trying to reply to and this is what I got 3d ago and now as well:

We're sorry, but your email message to ["discuss+1f2a66ede4c23843e7b27f6b218ee8fb@elastic.co"]
(titled Re: [Hadoop and Elasticsearch] Finding errors while inserting data into hbase throudh hive) didn't
work.

Basically it looks like the key provided in the email is not valid. Hopefully checking the records and potentially the email logs for this thread/user sheds some light.

17

And again, got an invalid reply to another thread:

We're sorry, but your email message to ["discuss+ea2de3fbd163a266c975d1d2828884e4@elastic.co"] 
(titled Re: [Hadoop and Elasticsearch] Is there a guide from creating uber jar with spark core and
elasticsearch-spark?) didn't work.

It looks at this point, at least for me, replying by email simply does not work...

Update: I sent two emails; the first one got the reply above, the next one made it as it can be seen in the thread here. This looks like a classical race condition when allocating keys...

18

+1ing this one. It happens to me very often when I reply via gmail. Very frustrating.

19

Today I spent a fair amount of time debugging this, first I added an option to very aggressively log any email failure

Once that was done I deployed and turned it on here, I have been trying to reply to this very topic by email and kept on getting that message.

When I looked at the log I saw nothing ...

So ... the process running this site were not even getting the messages, where were they going ??

Well, elastic is an enterprise customer, they have a staging instance with all data and settings, the staging instance has been polling email ... and it has reasonably old data.

I went ahead and disabled polling there, however... this leaves me wondering, do we even want to keep running this staging instance @nik9000 / @dadoonet are you using it for anything?

20

Haha! Funny thing. Thank you for debugging that!

I don't use myself the staging instance. I guess it was use before the official launch or to test some import scripts or so...

May be @drewr or @tyler would know?

21

Wow - well, at least you found the problem, regardless of whether we inflicted in on ourselves or not. :smile: Thanks for putting in the debugging effort, @Sam_Saffron.

We've used the staging instance here and there for things like testing style changes, piloting the Google Groups import, and so on, but we don't use it too often. If you anticipate more problems coming from the instance, then we should probably bring it down (if we can get another up relatively quickly if we need to test stuff.) If things look stable/safe to leave up, then I'd suggest leaving it up, as I'd hate to bring up a second staging instance for some reason in the future and have this problem duplicating if/when we forgot to change this setting again.

1 Like
22

Sure, no probs, I think we can leave it, just be careful with site settings on staging :slight_smile:

23

Yay! this seems to work for me now. Thank you!