SmarterMail Beta 7306 - In Production Real-World Experience So Far
Question asked by Shaun Peet - 1/6/2020 at 9:41 PM
We are running SM Enterprise Unlimited Mailboxes on a virtual server with a 2.10 GHz 6-core Xeon Silver CPU, and 8GB RAM running Windows Server 2012 R2.  We've got 524 domains and several thousand users.

I believe we're now on our 3rd Beta version of SmarterMail, after installing around Christmas (which wasn't the first Beta version out - probably around the 3rd version marked as production ready).  And yes, we are running this in production.

We've yet to witness any mission-critical errors.  We haven't officially notified our client base that MAPI is an option, but we run our own email on our mail server and I've tried MAPI on my own accounts - other than an issue with attachments which has been acknowledged and I'm sure will be fixed sooner than later (and doesn't appear to be a widespread issue), things are working pretty well so far using the basic features.  We will be letting a handful of our customers know that MAPI is now an option and letting them test it out this week as well.

As far as other protocols, EAS has done some odd things for me personally both in Outlook (before I converted to MAPI) and in mobile devices in that it sometimes will not sync non-email things (like contacts and calendar items) in both iOS and Android.  I don't know how widespread those EAS issues are but it appears there may be others in our world seeing strange things.  We don't have more than 30 of our thousands of users using EAS so this isn't a huge concern for us at the moment - and email syncing on EAS appears to be fine.

POP / IMAP are likely to be working just fine since nearly all our users are connecting via those protocols and we've yet to hear about any issues.

Very few people in our world use EWS and those who are out there haven't reported anything so I'd have to guess that's working normally too.

So, for everyone YMMV, but for us given the limited EAS exposure we're pretty pleased with the current production-ready Beta in all other protocols including MAPI.  The EAS issues we've faced aren't without viable work-arounds and may have some other fixes too, but for anyone with lots of EAS users on mobile I would proceed with caution.

As far as performance, we've yet to witness any significant improvements but there's also been no degradation.  Things were running a bit sluggish before I posted this and it appeared to be quite high RAM utilization on the server - possibly being caused by ClamAV.  After restarting the SmarterMail service things seem to be back up to speed but that's the first time I've had to do that since running the Beta.

I'd be curious to hear about anyone else who has been running in production.

7 Replies

Reply to Thread
echoDreamz Replied
You are lucky about POP and IMAP, we have hourly calls, chats and daily tickets for issues around it. Customers demanding updates and fixes. We are hoping ST has fixes soon. While some of the updates pushed out have fixed some of the NTLM issues, we are still having a good amount of clients complaining of Outlook issues and other items, IE, we had Outlook on macOS completely crashing, this has been fixed in a custom build provided to us.

If the POP3/IMAP issues were resolved, it would be perfection minus a few other issues. MAPI for me in my tests, worked great to download messages, but no longer downloads new messages.

EAS has always been wonky for some reason, many complaining of sync issues, showing unread items when a folder had none (mainly with iOS). Right now though, it seems like it is working, we are not receiving any reports of problems. ST always told us it was an iOS/EAS issue, but the client that was having the most problems, came from O365 where he said he didnt have the issue, eventually, he did cancel and move back to O365 where he happily emailed me about 4 months later saying the unread email count issue hadnt returned.

EWS on eM Client and Apple Mail on macOS, in my tests and others have been perfect, no issues that I can speak to at all.

I assume you do not use outbound gateways? Appears that the MAPI beta has an issue sending forwarded messages through the outbound gateways, normal emails seem to go just fine. If you enable them, all the forwarded messages just back up in the "Waiting for Delivery" queue and never leave.
Same issue in POP and IMAP as EchoDreamz here...
Gabriele Maoret - Head of SysAdmins at SERSIS
Currently manages 6 SmarterMail installations (1 in the cloud for SERSIS which provides services to a few hundred third-party email domains + 5 on-premise for customers who prefer to have their mail server in-house)
Matthew Titley Replied
Shaun, one 6-core and 8GB of memory works for thousands of users in your experience? Do you find that the web interface is reasonably fast with those specs? Do you mind sharing if your VMs are running on SSD or SAS/SATA? Only reason I ask is that before I make the jump to v17 I intend to roll out a new VM. I only have about 500 users and 30 domains, give or take. I was planning on assigning more "horsepower" than that even for my much smaller workload. Maybe my plans were overkill...


Shaun Peet Replied
With the caveat of *when SmarterMail is running as designed* (ie there's not a bug in the current version causing a degradation which would get fixed in a subsequent update), our VM is more than ample for our current workload.  The host machine is using all SSDs.

That being said, the other thing I'd caution is that workload is measured in different ways.  The number of inbound / outbound emails would also be a factor.  Our mail server is coupled with our own web application and due to that our mail server would see peaks when there's bad weather, but even with the peaks we've never seen an issue.  Here's the bandwidth overview report for the past few months:

Message Traffic:

Matthew Titley Replied
Thanks. Awfully kind of you to share the info.

Scarab Replied

Thank you for sharing the info.

We've been running v17 on a VM on 8 Cores and 24GB RAM for @ 270 domains and several thousand users and even though CPU usage has dropped dramatically in recent v17 updates from @ 99% to < 20% (Memory usage has stayed pretty constant at 44%) and although it is only half as bad as it was in earlier v17 releases we are still seeing unusually high I/O with 10-16 Current Disk Queue Length times and high %Disk Read/Write Time with almost 0 %Idle Time on a SAS Raid10 Array compared to v16 (and we only have a quarter of the Incoming & Outgoing traffic that you do...and offload Spam Filtering to our Incoming Gateways on top of that). I guess it may be time for us to consider upgrading to SSD Array.

Keeping fingers crossed that once the Beta goes Live there aren't POP/IMAP issues as some others are reporting. Glad to hear that it has been going well for you so far. (Knock on wood)
echoDreamz Replied
@Scarab we have around 20k users and our disk IO with the current beta build is really nice. We too are a RAID10 SAS array. We are 24 cores with 64GB of RAM.

Here is a 24hr sample of our disk queue length.

You can ignore the spikes at 12AM(ish) and 4:30AM(ish), those are related to our backups. 

Reply to Thread