[yukon]

Why does Apple require a restart with every software update? Point updates, iTunes, Safari, why restart? If the WindowServer is being updated by Safari (then not only is something very wrong) then just relaunch it. If there's a new kernel extension, unload the old and load in the new. The only time OS X needs to do a full restart is for upgrades to the core Mach kernel. I don't see how Safari should be updating Mach.

Why is this bad? Some people have network services on for example. If extension unloads cause problems, then those problems should be fixed, or we should move to a monolithic kernel and enjoy a large speed boost (UNIX benchmarks on Linux 2.4 vs. Mach 3.0 ). Maybe this is meant to encourage OS X Server use?

In MacOS X forum because it's not a limitation of Safari, nor very related to Safari anyway.

[gorickey]

I enjoy restarting once every month when updates like this occur...

[eevyl]

As it turns out, it is highly related to Safari.

Safari 1.2 update not only updates the application itself, but the system framework that actually renders the HTMland that it is sued by the OS and multiple apps, and so the OS needs to restart to apply those changes.

I guess a log out and back in is enough, but Apple forces the restart to be sure.

[yukon]

----geektalk
Sorry about the unqualified statement on linux vs. darwin. I was thinking of lmbench scores (basic kernel benchmarking), usually putting Linux 2.4 much ahead of Darwin. If you search google for "darwin lmbench linux", you get results with words like "order of magnitude worse", etc., and I haven't seen 2.6 benchmarks yet.

Here's a link comparing 10.2.1 with Linux 2.4.x
http://lists.terrasoftsolutions.com/...er/005547.html

Mach is designed in ways that cause larger delays, but for good reason. These differences allow kernel extensions to be loaded and unloaded dynamically, where Linux actually requires a recompile for most things.
----/geektalk

I guess my point is that if we have to restart for a simple 0.1 update to a browser, then we're throwing out the advantages to things we are already paying for.

[yukon]

Safari 1.2 update not only updates the application itself, but the system framework that actually renders the HTMland that it is sued by the OS and multiple apps, and so the OS needs to restart to apply those changes.

If WebCore is being updated, then Safari, Omniweb, and Help.app (?) should be restarted, not the entire system. A logout would do that, yeah, but even it isn't necessary.

A browser update shouldn't be touching the OS, but I'll grant it in this case (and don't say "Windows does it" ;-). In iTunes, it's updating the burner drivers (iTunes burns things with that service, almost makes sense...), and even that can be done manually by force-quitting iTunes installer, kextunload whateverdriver.kext && kextload whateverdriverupdate.kext.

[Horsepoo!!!]

Originally posted by yukon:
If WebCore is being updated, then Safari, Omniweb, and Help.app (?) should be restarted, not the entire system. A logout would do that, yeah, but even it isn't necessary.

A browser update shouldn't be touching the OS, but I'll grant it in this case (and don't say "Windows does it" ;-). In iTunes, it's updating the burner drivers (iTunes burns things with that service, almost makes sense...), and even that can be done manually by force-quitting iTunes, kextunload whateverdriver.kext && kextload whateverdriverupdate.kext.

Nobody cares about your uptime. Honest.

[dole]

If you really don't want to restart your computer after Safari is installed, just force quit the installer when it asks you to restart..

[yukon]

Nobody cares about your uptime

No kidding, neither do I. But uptime=availability. If I have to restart on every update, then my network goes down everytime Apple changes an icon.

[mitchell_pgh]

I know for a fact that you can don't need to restart, but it makes thing much easier simply to restart.

I'm cool with a restart every once and a while...

[Brass]

A restart every once in a while wouldn't be so bad, but I've just downloaded all the software updates from Apple that came out over the last week or so today: Airport, Java, Safari, Security Update.

On one machine I used Software Update to do them all, and that only requires one restart at the end.

However, on the next machine, I thought I would be a good net citizen and copy the downloaded packages from the first machine and install them manually.

Well, each of those 4 installs required a restart! That's 4 restarts in one day. Or if I was to have done them when they were first released, 4 restarts in one week.

So yes, once in a while would be okay, but this is a bit ridiculous for a unix based system with a package based installer. This simply should not be necessary.

No it's not a big deal to restart occasionally, but if you don't have to do it at all, surely that would be better???

If there was the possibility of creating these packages so that they could be installed without requiring a restart, then surely that should be done. I don't understand how some of you guys can argue that "I don't mind restarting so what's the problem?". If there is a choice between restarting, and not having to restart, then surely it would be better if we didn't have to? What's to argue about? Restart your machine if you want, but I'd rather not. No it's not a huge headache to have to restart my machine, but if if I don't have to then I'd rather not.

[Horsepoo!!!]

Originally posted by yukon:
No kidding, neither do I. But uptime=availability. If I have to restart on every update, then my network goes down everytime Apple changes an icon.

Well if being unavailable for the 45 seconds it takes to restart the computer is sooo important, you should start wondering why the heck you're updating things that aren't so important on that computer. You could have waited for a Security Update which really needs a restart and killed two birds with one stone.

[mrmister]

We've come a long way so that now people can have issues like this--in OS9 no one ever thought twice about it because we were restarting all the goddamn time.

[Rickster]

OmniWeb doesn't use the same copy of WebCore that Safari does -- we have a customized version of WebCore inside OmniWeb.app. Safari updates shouldn't affect OmniWeb unless they also include updates to the Cocoa frameworks themselves or lower-level OS components.

I can see Apple's rationale in "requiring" a restart for software updates that install system frameworks and such... it's a lot easier to explain that to a user than to tell them that they need to quit and restart everything (including OS components they may not be aware of) in order see the effects of the update everywhere. But it's still freakin' annoying.

[Brass]

Originally posted by Horsepoo!!!:
Well if being unavailable for the 45 seconds it takes to restart the computer is sooo important, you should start wondering why the heck you're updating things that aren't so important on that computer. You could have waited for a Security Update which really needs a restart and killed two birds with one stone.

It's not so much (for me at least) a matter of it being sooooooo important. But if it can be avoided, then it should be avoided.

It's details like this (ie, ease of use) for which many of us use Macs and not Wintels.

[wataru]

Originally posted by Brass:
Well, each of those 4 installs required a restart! That's 4 restarts in one day. Or if I was to have done them when they were first released, 4 restarts in one week.

In Panther you can install more than one .pkg at a time. I just did the Java and Safari updates at the same time (well, actually sequentially, but without a restart) on my girlfriend's machine.

[Gee4orce]

I'm pretty sure this is a support issue - ie. Apple don't want umpteen clueless users to ring in saying something doesn't work correctly because after doing an update they didn't quit and restart a particular combination of apps. Much easier just to force a restart.

[WJMoore]

Originally posted by yukon:
----geektalk
Sorry about the unqualified statement on linux vs. darwin. I was thinking of lmbench scores (basic kernel benchmarking), usually putting Linux 2.4 much ahead of Darwin. If you search google for "darwin lmbench linux", you get results with words like "order of magnitude worse", etc., and I haven't seen 2.6 benchmarks yet.

Here's a link comparing 10.2.1 with Linux 2.4.x
http://lists.terrasoftsolutions.com/...er/005547.html

Mach is designed in ways that cause larger delays, but for good reason. These differences allow kernel extensions to be loaded and unloaded dynamically, where Linux actually requires a recompile for most things.
----/geektalk

I guess my point is that if we have to restart for a simple 0.1 update to a browser, then we're throwing out the advantages to things we are already paying for.

While I agree with most of what you say, keep in mind the kernel used in Darwin is not a straight Mach microkernel. It is a hybrid, that aims to give the protection of a micro kernel with the performance of a monolythic kernel. As you say Linux and *BSD moreso often need recompiling to add or remove components. OS X can do this dynamically and still has performance hungry things (like filesystem drivers) inside the kernel, somehting a true microkernel wouldn't due. The follwing quote is from the Porting UNIX command line programs page on the ADC.

Although Mac OS X must credit BSD for most of the underlying levels of the operating system, Mac OS X also owes a major debt to Mach. The kernel is heavily influenced in its design philosophy by Carnegie Mellon's Mach project[17]. The kernel is not a pure micro-kernel implementation, since the address space is shared with the BSD portion of the kernel and the I/O Kit.

[Eriamjh]

Don't install updates and you won't have to restart.

[cybergoober]

Originally posted by Brass:
A restart every once in a while wouldn't be so bad, but I've just downloaded all the software updates from Apple that came out over the last week or so today: Airport, Java, Safari, Security Update.

On one machine I used Software Update to do them all, and that only requires one restart at the end.

However, on the next machine, I thought I would be a good net citizen and copy the downloaded packages from the first machine and install them manually.

Well, each of those 4 installs required a restart! That's 4 restarts in one day. Or if I was to have done them when they were first released, 4 restarts in one week.

So yes, once in a while would be okay, but this is a bit ridiculous for a unix based system with a package based installer. This simply should not be necessary.

No it's not a big deal to restart occasionally, but if you don't have to do it at all, surely that would be better???

If there was the possibility of creating these packages so that they could be installed without requiring a restart, then surely that should be done. I don't understand how some of you guys can argue that "I don't mind restarting so what's the problem?". If there is a choice between restarting, and not having to restart, then surely it would be better if we didn't have to? What's to argue about? Restart your machine if you want, but I'd rather not. No it's not a huge headache to have to restart my machine, but if if I don't have to then I'd rather not.

FYI -
(This only applies to Panther, AFAIK)
If you launch all the packages at once and get them going, they will all install. Each package will wait if it has to make changes to something that another package is updating. When they are all completed, there is ONE optimization and ONE restart…

Thought you might like to know that.

[Hop Pocket]

Gimme a break. Does it really bother you that much to reboot your machine? It is not as if these updates are coming every other hour.

Most people are really glad to have a Safari update that resolves their issues (especially banking). It bugs me that every update forces complaints like this out of the woodwork.

[Sir T Mann Esq.]

Originally posted by Horsepoo!!!:
Nobody cares about your uptime. Honest.

My Dual-1GHz Quicksilver (running the original release of Jaguar) now has an uptime of 467 days.

To be honest, I haven't used it at all for the past year or so (once I noticed that I was over 100 days, I was afraid I might crash it), so now I just leave it idle. Every day or so, I ssh from my TiBook to check the latest uptime stats.

My girlfriend's been trying to persuade me to [permanently] move in with her, but I just can't move out of my apartment... not after all that time! There must be some way of transporting the QS, preserving its uptime. Any ideas?


Tom

[Dale Sorel]

Originally posted by Hop Pocket:
It bugs me that every update forces complaints like this out of the woodwork.

Welcome to the Internet

[timmerk]

Stop your bitching. No one cares about your restarting problem. Go away.

[fizzlemynizzle]

Originally posted by yukon:
No kidding, neither do I. But uptime=availability. If I have to restart on every update, then my network goes down everytime Apple changes an icon.

If you cared about availability then your network's functionality wouldn't rest on one computer.

[bradoesch]

Originally posted by Sir T Mann Esq.:
My Dual-1GHz Quicksilver (running the original release of Jaguar) now has an uptime of 467 days.

To be honest, I haven't used it at all for the past year or so (once I noticed that I was over 100 days, I was afraid I might crash it), so now I just leave it idle. Every day or so, I ssh from my TiBook to check the latest uptime stats.

My girlfriend's been trying to persuade me to [permanently] move in with her, but I just can't move out of my apartment... not after all that time! There must be some way of transporting the QS, preserving its uptime. Any ideas?


Tom

Step 1: Send Dual 1 GHz to me.
Step 2: I'll try to beat your uptime.
Step 3: Profit. Erm, nevermind step 3.

[Axo1ot1]

Originally posted by Sir T Mann Esq.:
My girlfriend's been trying to persuade me to [permanently] move in with her, but I just can't move out of my apartment... not after all that time! There must be some way of transporting the QS, preserving its uptime. Any ideas?

Well if you are hard-line enough about it, she'll eventually dump you and you'll learn that there are more important things in life than silly computer stats.

[Moonray]

Apple should have enough brainpower to make an installation procedure that restarts just what's needed to restart. This restart for everything just to be sure becomes more and more microsoftish.

-

[diamondsw]

Originally posted by yukon:
No kidding, neither do I. But uptime=availability. If I have to restart on every update, then my network goes down everytime Apple changes an icon.

Gee, someone hasn't heard of maintenance windows... Restarts happen for a variety of reasons - that's why you schedule them.

Look, while I agree that restarts should be kept to a minimum, there is a major problem with your argument: WebKit can be used by third party apps. How is Apple supposed to know which apps to close so it can update them? What if you did something bizarre and used an app that embedded WebKit into the desktop and thus couldn't be quit? What if other users are using apps that use WebKit? You have to at the very least log out all users, and a restart is just a lot simpler.

[diamondsw]

Originally posted by Axo1ot1:
Well if you are hard-line enough about it, she'll eventually dump you and you'll learn that there are more important things in life than silly computer stats.

Actually, everything in life is more important than silly computer stats, especially THAT one.

[Cipher13]

Goddamn, a lot of you have no idea what you're talking about.

I totally agree with the original poster.

Imagine a real network server having to go down four times in a single week.

That is entirely unacceptable, even if for 45 seconds!

Unless a restart is ABSOLUTELY NECESSARY, no restart should be required by the installer.

For an end-user, more specifically, who really wants to have to close all their documents, and restart, interrupting their workflow terribly?

I know that every time I update a non-critical machine (my personal machine), I leave the restart for several days, with the installer app hidden (but not force quit, to remind me) until I've closed all of my stuff anyway.

It's annoying to have to do that.

[cybergoober]

Originally posted by Cipher13:
Goddamn, a lot of you have no idea what you're talking about.

I totally agree with the original poster.

Imagine a real network server having to go down four times in a single week.

That is entirely unacceptable, even if for 45 seconds!

Unless a restart is ABSOLUTELY NECESSARY, no restart should be required by the installer.

For an end-user, more specifically, who really wants to have to close all their documents, and restart, interrupting their workflow terribly?

I know that every time I update a non-critical machine (my personal machine), I leave the restart for several days, with the installer app hidden (but not force quit, to remind me) until I've closed all of my stuff anyway.

It's annoying to have to do that.

Well, if the machine is that effingham critical wouldn't it make more sense to wait for some off-peak time and run the four updates at the same time… requiring only one restart?

And why the effingham hell would a "real network server" need Safari updated?

On a personal machine, why is it not possible to wait until the end of the day or lunch, when workflow won't be interrupted, to install the update? Furthermore, if the machine is for production purposes, wouldn't it make more sense to hold off on the update for a day or two to see if there are widespread reports of problems with said update? Especially a non-critical update.

In short, if the update isn't necessary and you don't feel like having to restart, don't install it. Simple concept, really.

Deal…

[yukon]

With Windows, god only knows what's being installed, so every installer reccommends a restart, even many settings changes do (change your IP, for example). With MacOS, you only needed to restart if you had a new extension or system. With OS X, you only need to restart the system if you are literally replacing the most core componant of that system (XNU kernel itself).

As I said before, I was installing an update to the Safari browser, not to the OS (or at least I was told that). The Safari installer messed with the system, which could have broken programs that rely on WebCore (Omniweb being an example, except for the OmniGroup smartly taking care of this by including their own separate copy/version of WebCore/JSCore). This is the beginnings of an equivalent of "DLL Hell" on Windows, where installing simple programs remove, replace, upgrade, or install their preferred older versions of system components (breaking things relying on those components, like the system itself). When I installed a .1 iTunes update (3.2 or somesuch) a while back, it disabled support for my SCSI CDRW burner system-wide without any new features, would have skipped it if I had known, just an example that problems can happen this way on OS X the way I mentioned it does on Windows ("An MP3 player update broke my CD Burner").

If the system was being updated, then it should have been in the next system update. I thought a Safari update would be nice, and help when fixing my HTML's compatibility, and I skipped a few updates I knew Apple would force a restart on me. At the time, I remember joking with a friend online "installing a new Safari, heh just watch it make me restart, lol", but it did. I've force-quitted such things before, but that sometimes leaves an application's changes undone or unwritten (read: "Borks the update"). The fact that my friend was essentially installing a new GUI without restarting (new KDE version, full release out now though, hearing great things), that kinda drove the point home.

Since my Mac is stable and reliable, it provides network shares to my other machines. Taking the machine down because of a 0.1 browser update that sounded harmless, meant other, "crappier" systems had to restart. I had to stop what I was doing, save documents, close out, on more than one system. This was TOTALLY avoidable and unnecessary. OS X runs slower so that it is avoidable.

Sorry for the essay, a lot of people were misunderstanding me. I'm annoyed and worried, but I'm not furious or trolling.

[Developer]

Originally posted by yukon:
I had to stop what I was doing, save documents, close out, on more than one system. This was TOTALLY avoidable and unnecessary.

It told you beforehand it requires a restart. If it's uncomfortable to do at the moment, then update later.

[Cipher13]

Originally posted by cybergoober:
Well, if the machine is that effingham critical wouldn't it make more sense to wait for some off-peak time and run the four updates at the same time… requiring only one restart?

And why the effingham hell would a "real network server" need Safari updated?

On a personal machine, why is it not possible to wait until the end of the day or lunch, when workflow won't be interrupted, to install the update? Furthermore, if the machine is for production purposes, wouldn't it make more sense to hold off on the update for a day or two to see if there are widespread reports of problems with said update? Especially a non-critical update.

In short, if the update isn't necessary and you don't feel like having to restart, don't install it. Simple concept, really.

Deal…

Offpeak? Haha, yeah, right. 24 hours access is required. 45 seconds of downtime in a single day is unacceptable.

On a personal machine, I might have hundreds of documents/webpages open. To close them all to restart is a massive pain in the ass.

[cybergoober]

Then just don't update.



yukon - If you haven't noticed, when an update in Software Update requires a restart there is a little circle with a triangle in it to the left of the update. This indicates that a restart is required. Also at the end of the description of an update it will state, "restart required" if one is required. So it's not like the restart is sprung upon you with no notice. If you install an via a downloaded package, you will be notified via a sheet dialog that usually states something to the effect of, "Installing this update requires a restart…" with the option to cancel.


Those of you who make comparisons to UNIX/Linux are missing something here… Mac OS X IS NOT UNIX. It's not even that much based off of FreeBSD. They both just happen to share the same Mach microkernel.
As others have already stated, there are many reasons Apple has you restart rather than just killing the processes associated with the updated components.
If you can't accept that, then just don't install updates that require a restart until such time as you can deal with the "hassle".

If Apple didn't release these updates so often you same people would be bitching about how Apple needs to get on the ball and update some stuff.
Can't have it both ways…

[ginoledesma]

While I don't like the restart-required policy myself, at least its an improvement of what we used to have in the past, where certain installers would "require you to quit all applications" just to install an application.

I think most of us don't like being interrupted, or with being presented the slightest bit of inconvenience. For my part, I too wished that Apple would ask us to restart only if necessary. And for me, I'd like to have it limited to things like Mac OS updates, or major security updates.

For those who've used Linux, they know the wonders and benefits of not needing to restart while upgrading even major components of the system (like glibc). Its only an update to the kernel that requires a restart. Everything else can simply be "quit, updated, started." Minor inconvenience if any at all.

[JLL]

Originally posted by Cipher13:
Offpeak? Haha, yeah, right. 24 hours access is required. 45 seconds of downtime in a single day is unacceptable.

If you can't afford to have a server down for 45 seconds once in a while you should have more than one server.

I guess that you also have redundant power sources, redundant ISPs and so on, if you really require 100% uptime.

What does that server do since it needs to be a class 4 or 5 server? Running your nations health system?

[msuper69]

Originally posted by mrmister:
We've come a long way so that now people can have issues like this--in OS9 no one ever thought twice about it because we were restarting all the goddamn time.

Bingo! It's incredible how soon we forget the nightmare of constant restarts in OS9 (and earlier). Especially when most of the restarts were caused by an app crashing or heaven forbid, I moved the mouse, bringing down the whole OS.

Really, this is such a non-issue. At least we don't have to restart after EVERY SINGLE install of anything, like the Losedows OS requires.

Sorry but I just don't see the big deal here when the restarts can be managed very simply by just scheduling them properly.

[proton]

Originally posted by cybergoober:
Those of you who make comparisons to UNIX/Linux are missing something here� Mac OS X IS NOT UNIX. It's not even that much based off of FreeBSD. They both just happen to share the same Mach microkernel.

FreeBSD does not use the mach microkernel. FreeBSD uses its own kernel. The BSD personality for the mach kernel in OS X is based on FreeBSD, as are the userland commands. Mach is the part that didn't come from FreeBSD, it came from NeXTStep/OpenStep and through the NeXT lineage.

- proton

[cybergoober]

Cool. Learn something new every day.

The point I was trying to illustrate though is that Mac OS X is not UNIX…

[fizzlemynizzle]

Offpeak? Haha, yeah, right. 24 hours access is required. 45 seconds of downtime in a single day is unacceptable.

Then you would never have a single server for that application, you git. Learn what the hell you're talking about before you spout out of your ass again.

Signed,

Someone who does this for a living.

[yukon]

Didn't notice the "you need to restart" thing, it was a browser update (developer, iCab has never made me restart :-), I certainly wasn't expecting it. It isn't an "OH MY GOD! Think of the children!" issue, it's an "Oh, come on" issue.

Downtime is like a crash, and that should be unacceptable on the OS X Unix System. Yes, Unix. No, not UNIX, that's trademarked (by the Open Group probably) and would cost Apple unnecessarily. OS X is just as much of a Unix as Linux, some people use UNIX-Like system, for POSIX-compatible OSen (OSes, whatever). In fact, most would argue that OS X is more of a UNIX than Linux, as Linux intentionally tried to distance itself while OS X has roots in BSD (the kernel-tied compatiblity layer (oh boy, intellegent flames insue)). Apple has a graphic on their website, http://www.apple.com/macosx/features/unix/. OS X is "UNIX Based" in their phrasing (in order to evade the Open Group's wrath). http://www.unix.org/trademark.html. Community argues UNIX is now a generic term (like Xerox...uh, isn't? more applicably "Aspirin", thanks bayer) it's all a sordid tapistry, yea.

Anyway, my point remains, should you restart for a browser update? Should a browser update mess with the system (when a system update is forthcoming)? Since the hassle is entirely avoidable, and we've paid and continue to pay the price to make it avoidable, shouldn't we avoid it? If not, give me Copland.

[JLL]

Originally posted by yukon:
Anyway, my point remains, should you restart for a browser update? Should a browser update mess with the system (when a system update is forthcoming)?

It was a browser update AND a WebKit framework update.

[Cipher13]

Originally posted by JLL:
If you can't afford to have a server down for 45 seconds once in a while you should have more than one server.

I guess that you also have redundant power sources, redundant ISPs and so on, if you really require 100% uptime.

What does that server do since it needs to be a class 4 or 5 server? Running your nations health system?

There are about 15 servers, actually. Three of them being Macs.

Of course there are redundant power sources (UPS', as well as a generator. It is also on the local hospital power grid, so power rarely goes down). Two DSL lines, as well as the statewide frame relay system with its dedicated ISP (which is supposed to be primary, but we use it only as a backup given that the DSL is significantly faster).

45 seconds can mean a lot. We aren't afforded the luxury of offpeak hours - only when the Frame Relay goes down can we go down.

[ginoledesma]

Does Mac OS X Server follow the same restart-required policy for updates? I'd hate to think that it does, but I have the feeling it does anyway, considering it is a Server OS.

[JLL]

Originally posted by Cipher13:
There are about 15 servers, actually. Three of them being Macs.

Of course there are redundant power sources (UPS', as well as a generator. It is also on the local hospital power grid, so power rarely goes down). Two DSL lines, as well as the statewide frame relay system with its dedicated ISP (which is supposed to be primary, but we use it only as a backup given that the DSL is significantly faster).

45 seconds can mean a lot. We aren't afforded the luxury of offpeak hours - only when the Frame Relay goes down can we go down.

45 seconds mean absolutely nothing if you use load balancing, IP failover and so on. That way one server can shut down without anything happening to the service you offer.

[JLL]

Originally posted by ginoledesma:
Does Mac OS X Server follow the same restart-required policy for updates? I'd hate to think that it does, but I have the feeling it does anyway, considering it is a Server OS.

Considering it's a server you wouldn't normally install Safari updates (or Bluetooth updates for that matter).

[Twilly Spree]

Originally posted by JLL:
Considering it's a server you wouldn't normally install Safari updates (or Bluetooth updates for that matter).

He wasn't talking about Safari or BT updates. Just in general. You misunderstood.

[cybergoober]

Well not every update requires a restart… so what exactly is the question?

If the update touches frameworks and the such then yes a restart is required.

[dfiler]

Uptime concerns for a browser update?

If you can't afford 45 seconds of downtime... then that computer shouldn't be running a web browser. Or at least the admins shouldn't update a server's non-critical software unless it is already being taken down for maintenance.

Otherwise, I agree with the premise that some updates request an unnecessary restart... just not the contrived situation of a browser update causing server downtime.