Sync Gateway log: "setCheckpoint --> 409 Document update conflict"

Mobile Sync Gateway
1

I’m using Sync Gateway 2.6, CB Lite .Net 2.5.3 on Windows UWP

In the Sync Gateway log I see many info lines like this:
2019-10-22T20:36:22.697+02:00 [INF] SyncMsg: c:[789a1896] #7: Type:setCheckpoint --> 409 Document update conflict Time:449.237µs

At this time the app is running a push and a pull replication in parallel (not a PushAndPull replication).
Does the error indicate an update conflict when the checkpoint is saved ?
That shouldn’t happen, right ?
Or does the message indicate something else ?

I reviewed Document update conflict, it talks about the same log message, but I don’t understand @Jens response.

2

This is a conflict from a CBL replicator storing a checkpoint on the server. It’s not a document conflict in the usual sense.

This shouldn’t happen … does your client run multiple replicators at once with the same settings?

3

The app is running a Push and separately a Pull replication. Both are running at the same time, but that should be OK, because push and pull replicators use separate checkpoint docs, right ?

4

What is the configuration settings of the push and pull replicators ?

5

ReplicationConfiguration{
ReplicatorType = ReplicatorType.Push,
Continuous = false,
Authenticator = …,
PushFilter = (doc, flags) => DefaultPushFilter(doc?.ToDictionary(), IsDocFlagDeleted(flags)),
PullFilter = (doc, flags) => NoFilter(doc?.ToDictionary(), IsDocFlagDeleted(flags))
}

ReplicationConfiguration{
ReplicatorType = ReplicatorType.Pull,
Continuous = false,
Authenticator = …,
PushFilter = (doc, flags) => DefaultPushFilter(doc?.ToDictionary(), IsDocFlagDeleted(flags)),
PullFilter = (doc, flags) => NoFilter(doc?.ToDictionary(), IsDocFlagDeleted(flags))
}

private bool DefaultPushFilter(IDictionary<string, object> dict, bool isDeleted)
{
  if (isDeleted)
    return true;
  if (!dict.TryGetValue(nameof(IModelEntity.ReplicationType), out var dirObj))
    return false;
  if (!Enum.TryParse(dirObj.ToString(), out ReplicationDirection dir))
    return false;
  return dir == ReplicationDirection.Push 
              || dir == ReplicationDirection.PushPull 
              || dir == ReplicationDirection.PushThenPurge;
}

private bool NoFilter(IDictionary<string, object> dict, bool isDeleted) => true;

6

This may be a bug. I’ll have to check, but a separate push and pull replication may get the same checkpoint ID, which would explain the problem. If that’s true, I’ll file a bug report.

The CBL 2 replication protocol can push and pull over a single socket. So it’s assumed that you will run a single push/pull replication, rather than two replications. Is there a reason you need to run these separately? If not, you can fix the problem by switching to a single replicator.

7

Yup, this is a bug. I filed https://issues.couchbase.com/browse/CBL-501.

8

The design of our app framework originated when PushAndPull was not yet available.
I’ll explore making the switch from ‘Push and Pull separately’ to ‘PushAndPull’.
Thank you.

9

The design of our app framework originated when PushAndPull was not yet available.

Ah…from the 1.x days ? Because PushAndPull was available from day 1 on 2.x andit is the default mode if one isn’t specified.

10

Is this a SG bug or a CBL .NET specific bug?
I am using Android and iOS based CBL with SG 2.6 and one PushAndPull replicator. I have the same conflict message in the SG log. Is it the same bug or something else?

11

What is the message that you are seeing in the SG logs ?

12

I have thousands of this:

2019-12-16T08:02:27.157Z [INF] SyncMsg: c:[353b59de] #22: Type:setCheckpoint --> 409 Document update conflict Time:1.666115ms

13

Well, if you have a single replicator with pushandpull, I can’t see why you would see checkpoint conflicts (i.e. what is it conflicting with if there is only a single replicator with unique checkpoint) . Can you confirm you have a single replicator with pushAndPull.

You also mentioned you have iOS and Android app- did you happen to have a pre-built database file bundled with those apps ? If so, did you make sure to copy the database file over so they have unique UUIDs?

14

I confirm we have one PushAndPull replicator in Android and iOS. The Apps always start with an empty database

15

And that database is created from within the apps ? Is this a brand new deployment ? As in , were you setting up the clients and Sync Gateway for the first time ? When exactly did you start observing these errors ?

16

Yes the CBL database is created when a customer installs our App for the first time, then the App uses one PushAndPull replicator. We have few hundred channels, in each channel we have 1 to 5 devices, sometimes mixed Android and iOS

I am not quite sure what you mean with “were you setting up the clients and SG for the first time”?

I observe these error since weeks, we have them all the time.

Can these errors affect the replication? some of our customers regularly complain that some devices don’t get replicated for hours despite of having good internet connection.

17

Yes , it will affect replication - if you are observing this error, then your replication will keep restarting on every sync attempt because the checkpoint document is getting rejected due to conflict. Based on what you have shared above, I cannot determine how you can get into this scenario…why is the checkpoint doc in conflict.

If you have an Enterprise License, you can submit a support issue with relevant logs (sync gateway ) and code snippets (how CBL creates DB , sets up the replication etc) via our support channel. Also, can you upgrade to 2.7 and try out.

I am not quite sure what you mean with “were you setting up the clients and SG for the first time”?

Meant, if this was a brand new deployment or did this happen after an upgrade for instance and if this was something that happened from the start of your deployment or did something change in your environment that triggered this ?