What makes SmarterMail Enterprise an enterprise version? // Explanation of a big redundant setup and the problems we're facing.
Question asked by Falk Brockerhoff - January 7, 2015 at 5:11 AM
Unanswered
Hi,
 
we're running SmarterMail Enterprise in the unlimited version and activated EWS and ActiveSync for a little bit longer than a year. We faced some unsolvable problems and the one or other business impact, still unsolved. So I'm asking myself the question: what is the Enterprise word standing for?
 
We're running four MX Servers, external load balanced. Because of the lack of synchronizing the choice of the global admin for the domain greylisting state there are two MX servers building a redundant pair: one pair with activated greylisting and the other without. So the customer is able to choose if he wants greylisting or not by setting the right dans mx records. Why is this option not settable and synchronized via the web interface of the central Smartermail server?
 
Then we have three outgoing mail servers. But when we activate throttling for outgoing mails for a domain the throttling is done on the central Smartermail server, not on the outgoing gateways. Local delivery to the mailboxes for internal mails or mails received via the MX servers is delayed because of the huge mail queue, even if the main amount of these mails are for external delivery. Adding more outgoing mail gateways didn't scale for us.    
 
The three clamd-clamav and three spam assassin servers with linux are running fine and doing their job very good and fast. Didn't faced any problems here.
 
We want high availability, we need high availability, and our customers are asking for this. But the HA features of Smartermail are... well, how should I say? Not comfortable as we expected.
 
We set up two servers and installed Smartermail on each of them, both configured with their configuration and folder path on an UNC share. But only one of the two servers can be active at the same time - why not an active/active(/active/..) setup? In case of a failure we have to shut down the Smartermail service on the failing host properly and starting it at the standby node on our own (using a powershell script). For taking over the external reachable ip address we're using Microsoft NLB. The status of the primary Smartermail service has to be monitored on our own Smartermail isn't supporting Windows High Availability Services. For this we're using the software "Alchemy Eye" installed and running on the standby system.
 
Smarter mail even doesn't support Windows Storage Replica. That's why we're using two window server offering an UNC share, targeted and made high available with an external load balancer. File system content is replicated between this two nodes using the third party software Peersync. We were facing a bunch of problems until we find out the optimal settings and did some fine tuning. The Peer Software Support Desk was very helpful.
 
Beside the problems and compromises with setting up and keep Smartermail running, we faced some not fixable problems:
 
1.) EWS and ActiveSync is not running fine with Microsoft Outlook 2013. Many of our customer had several problems with it. Sometimes synchronization was not finished in time, recurring calendar events are running worse, mail replication (deleting mail in outlook was not synchronized to Smartermail) was done fragmentary. In short: we lost this customers, they are now using an original  Microsoft Hosted Exchange Service. This was not what we bought Smartermail for.
 
2.) Smartermail loses license information and is sometimes running in Free Edition Mode when doing a failover from primary server to standby system. We have to reset the activation history in our Smartertools Customer Portal and do activation of the active server manually to fix the problem. The time period it takes to alarm the technicians, getting him out of the bed, turning on his computer, dial into the vpn, getting the RDP session up and solve the problem, our mail services are completely not available. An unacceptable business impact. This issue was escalated to Smartermail support and on its way to be fixed soon, I hope.
 
3.) Since the upgrade to Smartermail 13 we're facing suddenly freezes of the Smartermail process, coming back after about 10 minutes. We're thinking that this problem occur when a customer is sending huge newsletters and we have some high traffic on outgoing mails. After processing thousands of outgoing mails for hours, the Smartermail process freezes. This issue is escalated to Smartermail Support Desk, too. But this ticket is open for over a month and we're still at the beginning of getting the Support Desk understanding and confirming our problem. We moved the concerned customer to a dedicated server running linux and postfix which is doing the job very well. No time to wait, because this issue is a business impact, too.
 
These are the main reasons why I'm asking me (and you): "for what stands the 'Enterprise' in 'Smartermail Enterprise'?". With the term 'Enterprise' I associate a very huge, very scalable and highly redundant and available setup with millions of users and mails getting processed. Can Smartermail beat the challenge?
 
 

10 Replies

Reply to Thread
0
Point 3) First update: in the meanwhile Smartertools Support was investigating Point 3 and came to the result that the reason of the issue is located at the fact that we are using an UNC share to store all the domain data. But a downgrade to Smartermail 12.x and sending a very huge newsletter without any impacts shows, that Version 13 of Smartermail is buggy. Keep you informed as soon as there are any news on this topic.
0
Point 3) Second update: this problem is (nearly) fixed with version 13.1.5498.27328, but we are still facing some freezes. Waiting for happening again and creating a core dump...
0
Point 2) First update: this bug should be solved in Version 13.1.5498.27328 and didn't happened again since the update. But we don't have any failover since then so that we can't verify this point.
0
Point 3) Third update: we installed another CustomBuild (Version 13.1.5504.29016) and did some tests with friendly users. The problems came back again, even worse than with version 13.1.5498.27328. We went back to version 12.5 again and won't do any updates for the next couple of weeks - until the situation calmed down for our customers and with the hope that the SmarterTools guys will fix the problem till then. We're out of this right now.
0
Point 2) Second update: with Point 3, Update 3 we did an update of the Smartermail software and switched to the standby server. Normally this is exact the situation where the license did get lost. But not this time, so I think that this issue is fixed right now. What a pity that SmarterMail 13.x is not stable enough for a production environment :-(
1
Steve Reid Replied
Smartermail is plenty stable for the average production environment.
 
Your situation is not the norm.
0
So.. Is stable the only enterprise feature for you? It's not what I need but if it came in the same package I would use it.
0
Steve Reid Replied
My comment was directly in reply to Falk's "SmarterMail 13.x is not stable enough for a production environment" I'm not really sure what you are referring to Richard.
0
Richard Frank Replied
Hi Steve, I thought you commented more to his fail over needs and HA these are nice enterprise features. I had some issues too but dev fixed that and than I could upgrade to the latest version.
0
Steve, hm, I'm a little bit disappointed: High Availability is not the norm? EWS / ActiveSyns is not the norm? Seriously? Should we better have staid with IceWarp Mailserver which is doing HA pretty well?!

Reply to Thread