runaway memory info

I have a verson of XP installed that has survived since before the product release for version 2.0. It is months old and has always worked perfectly and does not exhibit the memory leak problem. That is my baseline version (and so far, only version of XP).

This evening I did an XP install from iso image - the same one that I used for the baseline version. Immediately it began exhibiting the memory leak problem. I suspended the installation several times and restarted it - each time the memory would climb steadily and quickly. At 400meg I suspended it again (repeating several time until the install was completed). Had I not then surely all system memory would have been depleted. I finally arrived at a complete install and then added the Parallels tools. All the while the memory would climb. I was barely able to survive the tools installation as it had just about run the well dry when post-install reboot started. BTW, this is not a bootcamp install.

Running that version will cause memory usage to run away and if I permit it it would doubtlessly crash the system. The other instance continues to run with no change in memory usage or cpu usage changes.

Here is what I know - the earlier install is stable as a rock and always has been. I don't recall the release number of parallels then because I trashed it when the GA product came out. I have incrementally updated it as newer versions have come along and it's always worked perfectly. The fresh install just performed is completely unusable because of the memory problem. I would say that within 10 minutes of starting it, all memory will have been depleted.

 

have you gone through your parallels and vm prefs? when i went to rc1, i found a few surprises in the prefs. luckily i keep screenshots of previous builds before moving forward in anticipation of the possibility the prefs may be overwritten or unrecognized. i manually set memory (i give parallels 1024m - half of total, and i give the vm 640m). i noticed a 1.46g ".mem" file in /Users/Shared/Parallels but that file isn't persistent (flushes when i exit parallels).

don

 

The Parallels prefs memory is set for auto and the VM's for XP get 512 each. I'd noted the .mem file also and presume and also a .sav file which I suspect is either debug or on the path to snapshots ala VMWare. Speculation, purely. And I just know you're going to ask so yes, I've sent in what I have to the beta team

Being as it's 25 degrees outside and I'm on call anyway, I decided to muck with it a bit more. I cloned the XP instance I built tonight and have it and it's parent running side by side. The master has been growing but much slower (fewer resources to steal?) and the clone seems stable.

time marched on - it's been 20 minutes since I started this post. Both sessions have stabilized - one at 152 meg, the other at 138 meg of real memory, both have 2.14 g of virtual, and both are using about 10% of cpu.

Ten minutes on and both instances are now running IE on the msn.com page. Mem is at 151m each, cpu is 25% each. It would appear that the memory leak has stabilized. No changes were made to anything that would have caused the stabilization. The earlier problem of runaway memory has stopped on it's own. Computers don't work that way... Grrrr.

 

just curious, can you post a screenshot of "top -u 20"?

don

 

top -u 20
Processes: 74 total, 2 running, 72 sleeping... 237 threads 06:55:10
Load Avg: 0.84, 0.88, 0.83 CPU usage: 12.8% user, 33.6% sys, 53.5% idle
SharedLibs: num = 163, resident = 23.1M code, 3.46M data, 5.13M LinkEdit
MemRegions: num = 16430, resident = 1.29G + 11.9M private, 322M shared
PhysMem: 1.29G wired, 446M active, 255M inactive, 1.97G used, 26.2M free
VM: 15.9G + 109M 764735(0) pageins, 126987(0) pageouts

PID COMMAND %CPU TIME #TH #PRTS #MREGS RPRVT RSHRD RSIZE VSIZE
545 Parallels 32.7% 2:30:43 17 208 844 552M 133M 103M 2.15G
599 Parallels 11.1% 95:23.01 17 187 837 543M 127M 194M 2.13G
868 top 11.0% 0:03.91 1 18 20 524K 328K 992K 26.9M
0 kernel_tas 4.4% 13:59.52 48 2 1276 10.1M 0B 175M 1.36G
84 WindowServ 3.5% 6:43.08 3 594 7803 7.21M+ 113M+ 111M+ 828M+
354 Xquartz 2.3% 0:47.94 4 128 166 1.30M 9.01M 5.26M 366M
390 Activity M 2.3% 9:17.18 3 96 160 3.46M 9.70M 8.30M 371M
391 pmTool 2.1% 8:14.04 1 23 39 77.1M 792K 77.4M 125M
290 UniversalA 1.2% 0:28.47 1 64 108 592K 3.11M 1.20M 349M
376 VPNClient 0.7% 8:15.87 1 63 155 13.2M 10.3M 16.3M 377M
467 firefox-bi 0.4% 5:53.65 8 155 548 41.2M 28.8M 57.2M 512M
592 thunderbir 0.2% 2:26.83 6 95 462 39.6M 23.8M 52.5M 429M
362 cvpnd 0.0% 0:45.44 1 11 18 276K 944K 992K 28.3M
491 xterm 0.0% 0:00.16 1 14 75 496K 1.18M 1.25M 35.3M
289 Desktop Ma 0.0% 0:46.58 1 65 184 1.08M 5.37M 2.35M 356M
270 SystemUISe 0.0% 0:05.24 2 206 195 1.68M 6.10M 3.16M 359M

 

can you post a screenshot so the columns line up?

thx,
don

 

nevermind on the screenshot. the email notification uses monospaced fonts so the columns lined up. can you post a breakdown of your total installed ram, ram allocated to parallels desktop and ram allocated to each of the two parallels sessions you're running?

thanks,
don

 

Old_version is the XP install from months ago. Both sessions are running IE and the Outlook webmail client. Installed RAM is 2g.

Hardware Overview:

Machine Name: MacBook Pro 15"
Machine Model: MacBookPro1,1
Processor Name: Intel Core Duo
Processor Speed: 2 GHz
Number Of Processors: 1
Total Number Of Cores: 2
L2 Cache (per processor): 2 MB
Memory: 2 GB
Bus Speed: 667 MHz
Boot ROM Version: MBP11.0055.B08
SMC Version: 1.2f10

System Software Overview:

System Version: Mac OS X 10.4.8 (8L2127)
Kernel Version: Darwin 8.8.1
Boot Volume: Macintosh HD

This is dribbling in because I'm cooking breakfast - Parallels memory is set for auto and XP is given 512 meg each.

 

doesn't it seem like there's an issue with how parallels desktop is handling ram allotment when set to "auto"? you're running same setup as me...macbook with 2g ram. i haven't ran with two vm's at the same time but if two are given 512m each, given the overhead needed, that doesn't leave much for osx. 512m assigned to vm requires 20% more (just a guess)? what happens if you set memory for parallels desktop to 1024 (1/2 of your total installed ram) and 384 for each vm (so with overhead they don't require more than parallels desktop is given?

don

 

Parallels is now assigned 1024m, and each VM is given 384. The older install is on top and uses fewer resources.

 

can you post another output of "top -u 20"? i remember the second launched vm was using a lot more cpu than the first. i wonder if manually setting memory to parallels desktop makes a difference. i'm downtown in the wall street area now (deploying a raid for d2d2t) but i'm leaving soon so might be a while before i respond. not sure if i can offer any help, just following the steps i would take if i had the problem.

don

 

This is from a few minutes ago. Both VM's are doing the exact same thing (IE).