Force closing app company wide?

[HCHTech]

When we have a maintenance window scheduled to install updates for LOB apps, there is a typical requirement that no one be using that app. We send out a couple of reminder emails about getting out, but inevitably, someone (or a group of users) don't listen and leave that app open on their workstation, thwarting the update.

Right now because most of my clients are small, I typically remote into the offending workstations and close the app manually. I can't just close the open files from the server's computer management window as they will just open back up again. This process always adds time to the maintenance window.

I've considered scripting a task kill command to all workstations with our RMM - I could also force a reboot on the offending workstations, but that would likely result in lost data or corrupted files. I don't care about lost data - they were warned - but if it corrupts the apps database, then that's a bigger problem I'd like to avoid.

Is there an obvious solution I'm missing to make this inevitable job easier? Surely networks with 500 workstations have this same problem and don't do the one-at-a-time fix...

 

[mikeroq]

I'm sure it depends on the software. How it handles being terminated, or if it has a feature to remote logoff. For instance this law firm used Tabs3 software. Certain tasks in the program required all the other users logged out. You could "reset" a user but wasn't the best idea. The reset was mainly for if the program still said they were logged in, like say after a crash. So you would have to reset them to log in again. If I needed to do anything I basically had to make sure everyone logged out. I had to do this a few times so I could get in to export for their migration to a new system. The platinum version of the software says it has a feature to log out all users which would have been nice.

I imagine a LOB app that can support 500 users, might have such a feature.

 

[Blues]

I had a similar problem once before we had I think 5 satellite locations in different states and people left computers on and apps open all the time when they left for lunch or for the day. We would just send force reboot commands to the computers with maybe a 5 minute warning telling users to save work or it may be lost when PC reboots. Usually tacked in an extra note about any app that was being updated and to not use that app until a set time when expected completion would be.

 

[Sky-Knight]

I use the blunt force approach... order my RMM to reboot all the systems in question prior to my work getting started.

Users complain, but they also were notified in advance to get out, they're trained to log off at the end of each day and NOT leave crap running EVER, so management just reminds them of the above and we keep right on trucking.

Because you're right, you can't be spending the time it takes to do this all manually one machine at a time at that scale.

 

[britechguy]

I use the blunt force approach... order my RMM to reboot all the systems in question prior to my work getting started.

If (and I do mean if) you have the option to send a warning message like @Blues talked about, it's better to do that.

When I was a user, and certain situations required working odd-hours, it was very easy to forget "the regular schedule" because you just weren't normally in the office when everything was subject to a force shutdown. It's a courtesy to give a couple of minutes warning in case poor schlubs are toiling away so that they can save their work.

You still end up doing the reboot, but with at least a bit less upset. And end users have every right to be POed when work loss that could have been avoided with a polite 5-minute warning goes up in smoke because there wasn't one. There are human beings involved, and just a bit of consideration for those in "unusual circumstances" goes a long way toward making everyone happy without compromising your ability to get your job done.

 

[YeOldeStonecat]

For server based LOB app updates....I scheduled that ahead of time. I ensure that my primary contact at the client notifies everyone to close their programs at the end of the day.

I also can broadcast out a chat box via our RMMs agent. Pops right up in lower right corner!

When I get to work, it's after hours, so clients employees are gone for the day. If I get a warning from the LOB app that someone is still logged in...and it's a smaller client on a silver or gold plan, I'll log into that users workstation and close the app...and then reboot the server and begin my work on the LOB app upgrades. If it's a larger client...where it's impractical to go around to each workstation...I'll just reboot the workstations en masse...and then reboot the server. However the announcement to close apps and log off still goes out.

While it may be possible, I've never seen a database corrupt because a workstation connection to an LOB app got rudely severed. The end user might lose a little bit of their most recently entered data...but...they were warned.

Some LOB apps....from the server side, have a "disconnect clients" feature.

I do my best to accommodate clients as best I can, be polite, be patient. However...at the end of the day, since it IS the end of the day/into the evening when you're doing these things....our time is also precious and should be respected too!

About a month ago I was neck deep in these after hours upgrades with accounting clients. All those Thomson Reuters suites, or Intuit LaCerte suites.

 

[britechguy]

However...at the end of the day, since it IS the end of the day/into the evening when you're doing these things....our time is also precious and should be respected too!

Indeed. But you're doing exactly what I propose should be standard operating procedure. If there so happens to be a living human working after hours it allows them time to gracefully get out of whatever it is they are doing. If it's someone who walked away for the day with things still open, oh, well . . .

My concern is only that those poor souls who may be working after hours be given a chance to do a graceful exit. That's not too much to ask, and apparently, based on your own practices, you agree.

 

[HCHTech]

Good points, all. I just need to quit coddling them. It's January - start of a new year - the perfect time to discuss and implement changes to our SOP for maintenance windows. 2-day advance warnings will now announce a reboot is coming (and it's potential consequences if files are left open) instead of just requesting them to exit software and close any open files. Then a 30-minute advance warning of the reboot should cover the "never stops working" group.

 

[Sky-Knight]

Then a 30-minute advance warning of the reboot should cover the "never stops working" group.

There is nothing that can be done to placate these people. A good portion of them will simply get out of the way while being grumpy, others will insist you wait for them to finish something stupid important. I gave up... I just remind them that as announced, this process starts at X pm, if you're still in something it will be forcibly closed, your remote session will be terminated, do not pass go, do not collect $200.

Otherwise I'm up until 3 am just waiting for people to GTFO, and I'm too old for that crap. One site uses RDS for everything and I have to kill their VPN server or else they'll just reconnect in the middle of the upgrade and whine when things don't work.

 

[britechguy]

There is nothing that can be done to placate these people.

On that we can agree entirely.

But, it makes you look a lot better when a warning is issued close enough to the event that if someone is still working they can't say they were not warned.

I'm certainly not proposing, and never was, that you wait for these people to willingly GTFO. You have no way of knowing which are actual real people who may be "burning the midnight oil" versus the person who walked away at 5 PM. Everyone gets booted, but when you can say, "You were clearly warned," when someone screams it looks a lot better for you.

 

[YeOldeStonecat]

...

Otherwise I'm up until 3 am just waiting for people to GTFO, and I'm too old for that crap. One site uses RDS for everything and I have to kill their VPN server or else they'll just reconnect in the middle of the upgrade and whine when things don't work.

Yeah I'd do that for the larger clients....found out I'd just have to close the ports for RDG, or kill the VPN server service...so they wouldn't connect.

That was my larger client...had around 125 users at their peak. They're gone (hospital bought them out). Typical clients now are <30 users.