aberno@questor.wimsey.bc.ca (Anthony Berno) (03/05/91)
Well, after getting my cube+upgrade, I asked my salesperson why I didn't get an OD with system 2.0 on it (it was preloaded on the hard disk) and he explained that I had to buy it separately. After grumbling about having to use my own optical to back up the system, I posted a query on the net, to make sure that my salesman wasn't doing a snow job on me, and I was assured that, yes, you have to buy the OD separately and that I wasn't being snowed. Well... Today, my documentation arrived along with an OD containing system 2.0. But that is a matter of only slight interest. I have some more pressing queries: 1) When you add the size of root (found by using the Inspector panel in the workspace) with the amount of free space on the disk, is it supposed to total the size of the disk? I am short by about 80 meg on a 660 meg disk. No, I don't have a partition, in fact I removed it when I got my computer by doing a build disk. The transcript from the build revealed that it should have 690 meg, and when I opened up the computer to fix a little rattle, the sticker on the disk said it was 760 meg! Does anyone else have this problem with missing disk space? If you wouldn't mind, please do the above calculation and tell me what you get. I'm curious if this is normal - I suspect it is - but it bothers me to have an amount of space equal to 4 of the hard drives on my old Mac gone missing. '2) Also, is it a bad sign if a hard disk rattles? I opened up the machine and stuck a few pieces of cardboard around the disk drive, and it stopped, and I'm happy. Was it simply a jiggly mount, or is my drive going to die? Thanks very much to all past and future help. Believe me, I appreciate it! --- Anthony Berno (aberno@questor.wimsey.bc.ca) The QUESTOR Project: Free Public Access to Usenet & Internet in Vancouver, BC, Canada, at +1 604 681 0670.
scott@erick.gac.edu (Scott Hess) (03/05/91)
In article <B15ay2w164w@questor.wimsey.bc.ca> aberno@questor.wimsey.bc.ca (Anthony Berno) writes:
1) When you add the size of root (found by using the Inspector panel in the
workspace) with the amount of free space on the disk, is it supposed to
total the size of the disk? I am short by about 80 meg on a 660 meg disk.
No, I don't have a partition, in fact I removed it when I got my computer by
doing a build disk. The transcript from the build revealed that it should
have 690 meg, and when I opened up the computer to fix a little rattle, the
sticker on the disk said it was 760 meg!
Does anyone else have this problem with missing disk space? If you wouldn't
mind, please do the above calculation and tell me what you get. I'm curious
if this is normal - I suspect it is - but it bothers me to have an amount of
space equal to 4 of the hard drives on my old Mac gone missing.
A big culprit would be the version of du (or the du-like functionality)
provided by the Workspace Inspector panel. The default du(1) utility
provided gives the disk usage in bytes. This is misleading because
few files require an integral number of blocks and fragments - most
leave some free at the end of their frags. Thus, on average, if the
number of files in the system is n, the amount of data "lost" to
the space at the end of fragments is n*frag_size/2.
Of course, that's a simplistic answer. Most executable files, for
instance, use a multiple of 8k of space, and thus require a whole
number of fragments (actually blocks). On the other hand, large
files don't use fragments for the last couple k, and thus suffer
larger average data "losses".
Any way you put it, though, the more files you have, the more
space is lost to this phenomonon.
Anyhow, a solution to this particular problem is to grab the GNU
fileutil's version of du. It count's the size in real space used,
as in space allocated to the file (thus counting some unused space,
too). This should come closer to the real value.
Another place where data can be lost is due to the reserved blocks
on a disk which are reserved to help prevent fragmentation. BSD
FFS uses a weird and wonderful (with the emphasis on wonderful)
system to keep files from being spread across multiple cylinder
groups, and more importantly, to keep one file from forcing others
to be spread across multiple cylinders. To make this work well,
from 5 to 10% of the space on the file system is reserved. The
actual amount depends on whether it was optimized for space or
speed (that changes the way things work).
Lastly, there might be bad blocks and blocks used to replace them
counted in the disk's label, and those can't be used for regular
storage (well, it would sort of defeat the purpose, if they were).
Anyhow, more than likely all of your disk is there, somewhere . . .
Later,
--
scott hess scott@gac.edu
Independent NeXT Developer GAC Undergrad
<I still speak for nobody>
"Tried anarchy, once. Found it had too many constraints . . ."
"I smoke the nose Lucifer . . . Bannana, banna."
zimmer@calvin.stanford.edu (Andrew Zimmerman) (03/05/91)
In article <SCOTT.91Mar4235920@erick.gac.edu> scott@erick.gac.edu (Scott Hess) writes: >In article <B15ay2w164w@questor.wimsey.bc.ca> aberno@questor.wimsey.bc.ca (Anthony Berno) writes: > 1) When you add the size of root (found by using the Inspector panel in the > workspace) with the amount of free space on the disk, is it supposed to > total the size of the disk? I am short by about 80 meg on a 660 meg disk. >A big culprit would be the version of du (or the du-like functionality) >provided by the Workspace Inspector panel. The default du(1) utility >provided gives the disk usage in bytes. This is misleading because >few files require an integral number of blocks and fragments - most >leave some free at the end of their frags. Thus, on average, if the >number of files in the system is n, the amount of data "lost" to >the space at the end of fragments is n*frag_size/2. >Later, >-- >scott hess scott@gac.edu >Independent NeXT Developer GAC Undergrad While what Scott answered was true, there is another factor which might be causing some of the confusion. The unix fs reserves 10% of the size of the formatted disk to improve performance. This would be about 66 Megs on the 660 Meg disk. So, we have 790 - unformatted 660 - formatted 594 - unix file system The following is an attempt to be helpful. Since I have not tried this myself, please use at your own risk: "You can tune a fish, and you can tune a file system" or "How to nuke your filesystem" tunefs(8) Name tunefs - tune up an existing file system Syntax /etc/tunefs [ options ] Description The tunefs command is designed to change the dynamic parameters of a file system which affect the layout policies. The parame- ters which are to be changed are indicated by the options listed below: Options -m minfree This value specifies the percentage of space held back from normal users; the minimum free space threshold. The default value used is 10%. This value can be set to zero, however up to a factor of three in throughput will be lost over the performance obtained at a 10% threshold. Note that if the value is raised above the current usage level, users will be unable to allocate Restrictions This program should work on mounted and active file systems. Because the super-block is not kept in the buffer cache, the pro- gram will only take effect if it is run on dismounted file sys- tems. If run on the root file system, the system must be rebooted. I have left off all of the options other then the one that might have a direct effect on the original problem. Note: Don't blame me if something goes wrong. I have no reason to think that something will go wrong, I'm just being careful. Andrew zimmer@calvin.stanford.edu
rpruess@umaxc.weeg.uiowa.edu (Rex Pruess) (03/06/91)
In article <B15ay2w164w@questor.wimsey.bc.ca> aberno@questor.wimsey.bc.ca (Anthony Berno) writes: > Does anyone else have this problem with missing disk space? If you wouldn't > mind, please do the above calculation and tell me what you get. I'm curious Hmmm. I'm curious now. Looking at the mkfs and df output, there is a fair amount of space used before you even get a chance to save files on the disk. I assume mkfs takes a good chunk of space for overhead (e.g., superblocks, inode allocation, and reserved space for efficiency). Here's a chopped down version of the mkfs output: /etc/mkfs -N /dev/rsd0a 357560 22 10 8192 1024 32 10 60 4096 /dev/rsd0a: 357560 sectors in 1626 cylinders of 10 tracks, 22 sectors 366.1Mb in 51 cyl groups (32 c/g, 7.21Mb/g, 1728 i/g) The mkfs parm 357560 is the number of 1024 byte sectors. This yields a total space of: 357,560 * 1024 = 366,141,440 bytes The third to last mkfs parm is the minfree parm which is 10 in this case. Thus, 10% of the disk is reserved. See the mkfs man page for details. Output from the df command follows: % df Filesystem kbytes used avail capacity Mounted on /dev/sd0a 345711 225040 86099 72% / Note that used+avail is 311139 or about 10% less than kbytes (345711). This is due to the space reserved to allow the file system to work well. See the df man page for details. (I assume this 10% is a direct result of the mkfs minsiz parm.) The df output shows the partition has 354,008,064 bytes in it: 345711 * 1024 = 354,008,064 The difference between mkfs & df is about 3%. Again, I assume this is the overhead for superblocks, inode tables, etc. 366,141,440 (mkfs) - 354,008,064 (df) ----------- 12,133,376 (about 3% diff) -- Rex Pruess, Weeg Computing Center, Univ of Iowa, Iowa City, IA 52242 rpruess@umaxc.weeg.uiowa.edu (NeXTmail) (319) 335-5452
bennett@mp.cs.niu.edu (Scott Bennett) (03/06/91)
In article <RPRUESS.91Mar5104547@umaxc.weeg.uiowa.edu> rpruess@umaxc.weeg.uiowa.edu (Rex Pruess) writes: >In article <B15ay2w164w@questor.wimsey.bc.ca> aberno@questor.wimsey.bc.ca (Anthony Berno) writes: >> Does anyone else have this problem with missing disk space? If you wouldn't >> mind, please do the above calculation and tell me what you get. I'm curious > >Hmmm. I'm curious now. > >Looking at the mkfs and df output, there is a fair amount of space used >before you even get a chance to save files on the disk. I assume mkfs >takes a good chunk of space for overhead (e.g., superblocks, inode >allocation, and reserved space for efficiency). > >Here's a chopped down version of the mkfs output: > /etc/mkfs -N /dev/rsd0a 357560 22 10 8192 1024 32 10 60 4096 > /dev/rsd0a: 357560 sectors in 1626 cylinders of 10 tracks, 22 sectors > 366.1Mb in 51 cyl groups (32 c/g, 7.21Mb/g, 1728 i/g) > >The mkfs parm 357560 is the number of 1024 byte sectors. This yields >a total space of: > 357,560 * 1024 = 366,141,440 bytes Worse yet, many (if not most) disk drive vendors incorrectly advertise the capacity of their products. For example, in the above case many vendors would advertise the drive as a 366MB drive. This is wrong because 1MB=1024KB=1048576B. So, from this we can calculate in reverse that 366141440B=357560KB=~349.18MB, *not* 366MB. It would be nice if the vendors would stick to straight info, rather than just adding to the confusion. > > [remainder of text deleted --SJB] >-- >Rex Pruess, Weeg Computing Center, Univ of Iowa, Iowa City, IA 52242 >rpruess@umaxc.weeg.uiowa.edu (NeXTmail) (319) 335-5452 Scott Bennett, Comm. ASMELG, CFIAG Systems Programming Northern Illinois University DeKalb, Illinois 60115 ********************************************************************** * Internet: bennett@cs.niu.edu * * BITNET: A01SJB1@NIU * *--------------------------------------------------------------------* * "I, however, place economy among the first and most important * * of republican virtues, and public debt as the greatest of the * * dangers to be feared." --Thomas Jefferson * **********************************************************************
marcus@eecs.cs.pdx.edu (Marcus Daniels) (03/23/91)
Is 20 meg definitely worth it? What is the best way to watch swapping activity? Edit doesn't seem to search backwards correctly with regular expressions. Anyone using epoch? ------------------------------------------------------------------------------ - marcus@eecs.ee.pdx.edu / ....!uunet!tektronix!psueea!eecs!marcus - "The power of accurate observation is called cynicism by those who don't have it"
fischer@iesd.auc.dk (Lars P. Fischer) (03/25/91)
>>>>> On 22 Mar 91 23:50:45 GMT, marcus@eecs.cs.pdx.edu (Marcus Daniels) said:
Marcus> Is 20 meg definitely worth it?
Most Unix workstations run at 10-15% of their real performance, due to
too little RAM.
Marcus> What is the best way to watch swapping activity?
Try "man vmstat".
/Lars
--
Lars Fischer, fischer@iesd.auc.dk |Erst kommt das Fressen, dann die Moral
CS Dept., Univ. of Aalborg, DENMARK. | - B. Brecht
madler@pooh.caltech.edu (Mark Adler) (03/25/91)
In article <FISCHER.91Mar24231722@galilei.iesd.auc.dk> fischer@iesd.auc.dk (Lars P. Fischer) writes: >Try "man vmstat". Actually, it's called vm_stat instead of vmstat on the NeXT. And if you don't have the extended release, obviously man vm_stat won't help much either. Just do vm_stat. The output is pretty much self-explanatory, if you're familiar with virtual memory techniques. Mark Adler madler@pooh.caltech.edu
zazula@uazhe0.physics.arizona.edu (RALPH ZAZULA) (03/26/91)
In article <FISCHER.91Mar24231722@galilei.iesd.auc.dk>, fischer@iesd.auc.dk (Lars P. Fischer) writes... >>>>>> On 22 Mar 91 23:50:45 GMT, marcus@eecs.cs.pdx.edu (Marcus Daniels) said: > >Marcus> Is 20 meg definitely worth it? > >Most Unix workstations run at 10-15% of their real performance, due to >too little RAM. > >Marcus> What is the best way to watch swapping activity? > >Try "man vmstat". > isn't it "man vm_stat"? I think so since the command is vm_stat. Also, the program Monitor (available at most NeXT archive sites in the 1.0-release directory) shows a realtime display of swapping activity. >/Lars >-- >Lars Fischer, fischer@iesd.auc.dk |Erst kommt das Fressen, dann die Moral >CS Dept., Univ. of Aalborg, DENMARK. | - B. Brecht Ralph |----------------------------------------------------------------------| | Ralph Zazula "Computer Addict!" | | University of Arizona --- Department of Physics | | UAZHEP::ZAZULA (DecNet/HEPNet) | | zazula@uazhe0.physics.arizona.edu (Internet) | |----------------------------------------------------------------------| | "You can twist perceptions, reality won't budge." - Neil Peart | |----------------------------------------------------------------------|
cs00jec@unccvax.uncc.edu (Jim Cain) (04/29/91)
I will be losing access to the net, ftp, etc. in the next week when my account is deleted at the end of the semester. I have been ftp'ing stuff like mad -- about 20-25MB over the last two days (this is not fun at 1200 baud!! But that's the only speed available on our school's network). I have a few final questions to ask before I can no longer ask them or receive the answers: 1. UUCP: The school "does not have the resources" to give me UUCP access, so I must look elsewhere. Portal Communications offers UUCP access for $39.95 per month plus $1.95 per hour. For tech support it is $95.95 (!) per month. Is this reasonable? Also, they offer the complete cnews package, including documentation, for $595. I know I can ftp this stuff, but with no support or documentation, what are my chances of getting it set up? I know a good bit about UNIX but have little experience with UUCP. What about Nutshell's book _Managing UUCP_? Is this enough to get it all set up? Alternatively, for $10 a month (plus $19.95 to open), Portal will give you a login to read news and mail from their system. This is much cheaper, but I'd rather do all this from NeXTstep. Some of the stuff I've downloaded are all the cnews sources, trn and rn from uunet; and NewsGrazer from purdue. (BTW, Grazer is fantastic!) Once I get the extended release, will I be able to compile this C code? Are there any libraries, etc. I should ftp now? 2. I have noticed that when you Save in a Print... dialog, the resulting PostScript does not include the fonts used. Is there any way to have them included? I work at a service bureau, and this is a common way to have documents printed, especially from PC's since we do not support any PC applications. Without font data, the printer must have (an identical version) of the font already downloaded. I can't print my docs on our Lino's unless the fonts are included. 3. Does the extended release include all the sources/binaries/etc. necessary to recompile the kernel? I would like to maintain disk quotas on my NeXT for my roommates/friends who have logins. 4. Is there any good BBS software out there for UNIX or NeXTstep? A NeXTstep BBS sounds like a good project. Maybe I'll do it once I buy another hard disk and get the extended release installed. 5. I recently saw references to UPS's but cannot find them again. Does anyone have any recommendations on manufacturers/models/dealers? Well, that's all I can think of for now. Many thanks for any info. Jim -- =================== Jim Cain * cs00jec@unccvax.uncc.edu =================== The University of North Carolina at Charlotte College of Engineering * Dept of Computer Science ========== "'ave you been shopping?" * "No, I've been shopping." ==========