Dopus informed me this morning that an update from 11.3 to 11.4 was available, so I let it download/install the update.
The Installshield wizard starts, I get the welcome screen and the license agreement, then shortly after that screen I get the 'has stopped working' screen.
Problem details are as listed below.
Windows 8.1 64-bit, i7-4770k not overclocked, 32Gb ram, GTX 760, drivers current.
Doesn't help:
Rebooting, closing all possible apps first, running as admin, clearing temp, halting Kaspersky.
Comparing automatically downloaded installer to manually downloaded on shows no binary differences, and both copies of installer exhibit same behavior.
Haven't tried:
Changing DEP from 'for essential windows programs and services only' to 'turn on DEP for all except those I select'.
Would rather avoid that due to problems it causes with other software, and haven't had to use it for previous updates to dopus (only 10 and 11 have been installed on this particular pc as it is only a few months old).
Suggestions?
Problem signature:
Problem Event Name: BEX
Application Name: setup.exe_Directory Opus
Application Version: 11.4.0.0
Application Timestamp: 4e51d676
Fault Module Name: StackHash_e98d
Fault Module Version: 0.0.0.0
Fault Module Timestamp: 00000000
Exception Offset: PCH_41_FROM_ntdll+0x0003F521
Exception Code: c0000005
Exception Data: 00000008 OS Version: 6.3.9600.2.0.0.256.48
Locale ID: 1033
Additional Information 1: e98d
Additional Information 2: e98dfca8bcf81bc1740adb135579ad53
Additional Information 3: 6eab
Additional Information 4: 6eabdd9e0dc94904be3b39a1c0583635
Check the installer exe you downloaded has a good digital signature, in case the file is corrupt (the same file may have been downloaded both times due to e.g. transparent web caching):
Failing that, it is unlikely to be DEP as all our machines have the default DEP setting or more strict settings (my main workstation has DEP enabled for everything) without issues.
If the file is not corrupt/modified then I would look at things unusual to your setup, perhaps a tool which hooks into other programs to change their behaviour, which may be loaded into the installer and going wrong. We have not had any other similar reports in the 21 days since 11.4 was released.
I managed to resolve the issue, than you for the response Leo.
I did try changing DEP, but that did not resolve it as you suspected.
The resolution was to point the Windows temp system variables to a physical drive instead of a ram disk.
As soon as I did that (no reboot necessary) the installation worked fine.
The interesting thing is I have been using a ram disk in the same manner all along on this pc; that's part of the reason I have 32gb ram on it.
The only thing that changed was about a month ago I changed the temp files ram disk from NTFS to FAT32 to test what, if any, speed differences that made.
With previous opus updates that succeeded, the temp files would have been on an NTFS ram disk; this time they were on a FAT32 ram disk.
I am not aware of any other issues I have seen with using a FAT32 ram disk (besides opus updates throwing a BEX event error).
Anyway in testing I didn't see enough of a speed difference to make the FAT32 ram disk worthwhile, so I just changed it back to NTFS.
Many ram disk products default to FAT32 (some even FAT16), so this problem might be seen again by somebody else.
With decreasing cost of RAM and increased use of SSDs, ram disks are becoming somewhat popular again.
If you'd like to test this, here are the settings I used and had the problems with:
Product: SoftPerfect RAM Disk 3.4.5 64-bit
Size: 4096 MB
File system: FAT32
No compression
Image file saved to a physical drive and restored at boot
No problems with previous dopus updates with ram disk same as above except NTFS instead of RAM32.
Anyway, thanks for a great product which is an indispensable part of my day.
Mark.
I forgot to mention -- all four 'TEMP' and 'TMP' environment variables normally point to a couple of folders on the RAM disk.
To test if that is the problem or not I set all 4 variables to folders on my D: drive, which is a 2Tb hard drive, and then retried the installation immediately (no reboot) and it worked fine.
I then deleted the ram drive, recreated it as NTFS, and repointed temp variables to it.
In retrospect I should have temporarily pointed the temp variables to an NTFS ram drive and then retried the installation, as that would have been a better test of "is it because it is a ram drive or is it because of FAT32".
Since previous updates worked with an NTFS ram drive I suspect the FAT32 is the culprit, but I'll find that out with the next dopus update.
I'm not sure what you mean by that -- using a Ramdisk for temp and cache storage makes a huge difference in a lot of the activities I do -- for example rendering or building a large software project on a ram disk is much faster than on an SSD, and many times faster than on a hard drive.... I'll continue to make changes to my system that are needed to get my work done in a timely manner, like I have since the early 80's when I first started working in IT. When I find something that might make an efficiency difference in my workday I will always test it out and weigh the pros and cons.
In this case I see a ram disk as a very helpful tool, and haven't seen any problems in using one formatted as NTFS. In a month of fairly heavy testing with one formatted as FAT32 the only problem I saw was with the dopus installer. I did some further experimentation after posting a few hours ago. It looks like the 64-bit dopus installer will fail on updates if temp storage points to a FAT32 drive but not if it points to an NTFS drive. It appears the failure occurs when the installer is trying to stop/remove the existing dopus (specifically when it is trying to stop it). If you have temp storage on a FAT32 drive but manually uninstall the existing dopus before running the new installer, there are no problems encountered.
I just thought you might be interested in the cause of the issue in case another user had the same problem, especially since Fat32 is the default for a lot of ram disk software and their popularity is currently increasing.
What were the performance improvements you measured after switching from NTFS to FAT for your temp dir? (Windows requires the system drive to be NTFS (or at least not FAT). Using anything else for the system or temp drives is asking for trouble.)
What other issues are going to happen by having a temp dir that does not survive reboots, as some software will assume it will? (Files placed in temp are not guaranteed to last forever, but they can be assumed to last through a reboot, and installers do take advantage of that for post-reboot registration or clean-up tasks.)
Is any of this worth it for so little benefit? That was my point.
(Also, just because the temp dir is on disk does not mean the files are actually being written to disk. With enough RAM, and files opened with the right flags, the files will never actually touch the disk.)
I find this confusing. The thread is talking about a ram disk, not a physical hard drive or ssd, so of course it is not going to be the Windows system drive?
As stated in the above posts, FAT is the default in a lot of ram disk software, so a lot of users of ram disks are going to have them as FAT rather than NTFS. In fact, some ram disk software doesn't give the option of using NTFS. For the past few years I have been using a commercial ram disk that does support NTFS, and NTFS is what I have been using.
I got curious about the speed differences between NTFS and FAT for ram disks < 4gb, so I tested that for about a month. CrystalDiskMark showed roughly a 15% speed increase in random writes for a FAT ram disk vs an NTFS ram disk, less but measurable increases for reads and sequential writes. For real world tests I timed batch processing in Photoshop, builds of large c++ and c# solutions in Visual Studio, and rendering in 3ds Max, and also just subjectively used it for a month. The timed processes saw a 5% - 15% speed improvement for FAT ram disk over NTFS ram disk. The subjective daily tests showed no real difference that was noticeable in daily operations, but also no problems until I tried to update dopus.
Speed improvements for a ram disk over a hard drive or ssd are much larger than the difference between different formats of a ram disk. I won't go into benchmarks as there are plenty of websites that do that. For the activities I do daily the use of a ram disk is a no-brainer, but for the average pc user it is probably not worthwhile. Ram disks are created by and bundled with the motherboards of several manufacturers (ASRock and ASUS for example ), (AMD ram ), some SSDs and other hardware. I don't personally feel it is an issue, but a lot of people use ram disks because the perception is writing temp files, browser caches, etc. to an SSD will wear it out rapidly (that's why ram disks are sometimes bundled with SSDs). There are a lot of people that use a ram disk for the windows page file. I (and others) have tried to explain to people why that is a bad idea, but the practice persists.
Most if not all modern ram disks give you the option of creating the disk with each boot or persisting an image. Some compress the image, some don't. Some do binary updates to the image as changes occur or timed interval updates during idle time. Some rewrite the image as the computer is shut down. The general recommendation is to use an image if the purpose of the ram disk is system temp files or other items that should persist between reboots; the ram disk service is one of the first things to load on boot, and it builds the drive from the image before any other software (including Windows) try to access temp or cache storage.
Does it make a difference in boot and shut down time to save the image? Yes, definitely. It means shut down takes a few seconds longer and boot takes a few seconds longer. If 5-10 seconds at boot time are really important to a user then a persistent ram disk probably isn't a good option, but most people spend a heck of a lot more time USING their computers than REBOOTING them. What I've found over the years works best for me is to have two ram disks with one being volatile and one persisted, with a scheduled cleaning of temp and cache space so they don't become too large.
Again, not sure what you mean.
[ul]
[li]Are you asking if a ram disk is worthwhile? It depends on what you use your computer for and how much ram is in it. In my case it means builds, rendering, and several other activities take a fraction of the time they would take even if done on an SSD, so I wouldn't call it 'little benefit'.[/li]
[li]Are you asking if FAT32 is worthwhile over NTFS for ram disks? As previously stated, FAT32 is the default for many ram disks, and the only option in some. I have been using NTFS for ram disks but tested using FAT32 for a month or so. I saw a measurable speed increase with FAT32, but the speed difference between FAT32 and NTFS is a tiny fraction of the speed increase of ram disk in general over ssd. I decided to go back to NTFS after seeing the instability in the dopus installer trying to shut down the previously installed instance, but again that is the only problem I saw in a month of use.[/li]
[li]Are you asking if FAT32 is worthwhile over NTFS for temp and cache storage on hard disks? I wouldn't know as I haven't tested that and don't plan to since there are other much better options available for increasing productivity.[/li][/ul]
I don't see anywhere on the dopus website where an installer is available for download that has special switches to use RAM instead of temp space. I guess if a user really wanted to he could research each different executable and each flavor of installer to see if there are switches available to force temp alternatives, then build dos .bat or .cmd files or powershell .ps1 files for the various purposes. It seems like that would be counter-productive as more cycles would be spent in the activity than what is gained by performing it. Personally I'd rather have a ram disk V: that is for general purpose use and a ram disk W: that is for temp and caching, point downloads to V:, and then not have to do anything but double-click on the download to enjoy the extra speed. But to each his own!
Anyway... it doesn't seem this thread is trending in a positive direction, so I won't post more in it.
I had an issue with dopus upgrade, figured out the solution, then did additional tests and posted the results here hoping they would be beneficial to dopus support folks or other users seeing the same issue.
Ram disks are becoming more pervasive and their use is increasing. Many ram disks have installation options to automatically move temp storage and browser caches to ram disk, so it is right there in people's faces at installation time. Since FAT32 is the default with some ram disk software, I thought the information might be useful here.
