Publish occasionally results in a duplicate sessionId trying to be added

[mizzao]

Rare error causes strange results. Needs tests added and fix.

See also

• http://stackoverflow.com/questions/17865319/strange-meteor-error
• https://groups.google.com/forum/#!topic/meteor-talk/r6ENlLsrxoY
• Inserting multiple items with the same _id into a local collection results in very confusing error meteor/meteor#1246

[tanis2000]

It happens to me as well, with this kind of error:

I2049-18:23:51.248(1)? qiB8NrKQ7figAqWt8 with session R2KkLCmp3yQsATM2n logged out
I2049-18:23:51.872(1)? qiB8NrKQ7figAqWt8 with session KQszzdRiJqSCevNLn logged in from 127.0.0.1
I2049-18:23:51.886(1)? Exception from sub undefined MinimongoError: Duplicate _id 'KQszzdRiJqSCevNLn'
I2049-18:23:51.886(1)?     at MinimongoError (packages/minimongo/minimongo.js:46)
I2049-18:23:51.886(1)?     at LocalCollection.insert (packages/minimongo/minimongo.js:460)
I2049-18:23:51.887(1)?     at Meteor.Collection.(anonymous function) [as insert] (packages/mongo-livedata/collection.js:448)
I2049-18:23:51.889(1)?     at Package [as _handler] (packages/user-status/user_status.coffee:39:14)
I2049-18:23:51.889(1)?     at maybeAuditArgumentChecks (packages/livedata/livedata_server.js:1356)
I2049-18:23:51.889(1)?     at _.extend._runHandler (packages/livedata/livedata_server.js:777)
I2049-18:23:51.889(1)?     at _.extend._startSubscription (packages/livedata/livedata_server.js:675)
I2049-18:23:51.889(1)?     at packages/livedata/livedata_server.js:352
I2049-18:23:51.890(1)?     at Array.forEach (native)
I2049-18:23:51.890(1)?     at Function._.each._.forEach (packages/underscore/underscore.js:79)

[jonjamz]

Have you thought about using an upsert? Seems like this might be a case for that.

[mizzao]

@tanis2000 do you want to see if this is still happening on the latest version (0.1.6)? I added @jonjamz' upsert code. Also, were you able to replicate this error consistently before?

[tanis2000]

It happens quite often but I don't know exactly why. I'll try the new version and I'll let you know.

Valerio Santinelli

Inviato da iPhone

Il giorno 04/dic/2013, alle ore 17:06, Andrew Mao notifications@github.com ha scritto:

@tanis2000 do you want to see if this is still happening on the latest version (0.1.6)? I added @jonjamz' upsert code. Also, were you able to replicate this error consistently before?

—
Reply to this email directly or view it on GitHub.

[mizzao]

@tanis2000 Whoops, you'll have to actually use 0.1.7. One must be very careful with using upsert properly. 0.1.6 doesn't track users properly.