* Why minor is still 8 bit? @ 2012-06-04 1:25 George Shuklin 2012-06-04 1:35 ` Al Viro 0 siblings, 1 reply; 11+ messages in thread From: George Shuklin @ 2012-06-04 1:25 UTC (permalink / raw) To: linux-kernel I've understand that major and minor numbers for device is 8-bit size. I've don't understand, WHY? In 2012! (I've just have some discussion about 8-bit counter limitations in context of cloud computing. In middle of 2012!!!!) But seriously, I've done some googling but found no answer why it still just 8 bit size. Is this some serious backward compatibility? Or just there is no one with itch strong enough? ^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: Why minor is still 8 bit? 2012-06-04 1:25 Why minor is still 8 bit? George Shuklin @ 2012-06-04 1:35 ` Al Viro 2012-06-04 3:53 ` George Shuklin 0 siblings, 1 reply; 11+ messages in thread From: Al Viro @ 2012-06-04 1:35 UTC (permalink / raw) To: George Shuklin; +Cc: linux-kernel On Mon, Jun 04, 2012 at 05:25:01AM +0400, George Shuklin wrote: > I've understand that major and minor numbers for device is 8-bit size. > > I've don't understand, WHY? In 2012! > > (I've just have some discussion about 8-bit counter limitations in > context of cloud computing. In middle of 2012!!!!) Yes, shocking. It's been several years and you are still using that buzzword; I mean, it's _so_ 2010... Should've moved on to whatever's in this year... Incidentally, minors are 20 bit and majors are 12, but don't let that stand in the way of righteous indignation - what's mere facts when one is having discussions in context, presumably leveraging synergies all along... ^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: Why minor is still 8 bit? 2012-06-04 1:35 ` Al Viro @ 2012-06-04 3:53 ` George Shuklin 2012-06-04 16:11 ` Casey Schaufler 0 siblings, 1 reply; 11+ messages in thread From: George Shuklin @ 2012-06-04 3:53 UTC (permalink / raw) To: Al Viro; +Cc: linux-kernel On 04.06.2012 05:35, Al Viro wrote: > On Mon, Jun 04, 2012 at 05:25:01AM +0400, George Shuklin wrote: >> I've understand that major and minor numbers for device is 8-bit size. >> >> I've don't understand, WHY? In 2012! >> >> (I've just have some discussion about 8-bit counter limitations in >> context of cloud computing. In middle of 2012!!!!) > Yes, shocking. It's been several years and you are still using that > buzzword; I mean, it's _so_ 2010... Should've moved on to whatever's > in this year... > > Incidentally, minors are 20 bit and majors are 12, but don't let that > stand in the way of righteous indignation - what's mere facts when > one is having discussions in context, presumably leveraging synergies > all along... Ok, thank you for information. But question is almost same: can I have 30k active logical volumes on my single server, please? ^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: Why minor is still 8 bit? 2012-06-04 3:53 ` George Shuklin @ 2012-06-04 16:11 ` Casey Schaufler 2012-06-04 18:59 ` George Shuklin 0 siblings, 1 reply; 11+ messages in thread From: Casey Schaufler @ 2012-06-04 16:11 UTC (permalink / raw) To: George Shuklin; +Cc: Al Viro, linux-kernel, Casey Schaufler On 6/3/2012 8:53 PM, George Shuklin wrote: > > On 04.06.2012 05:35, Al Viro wrote: >> On Mon, Jun 04, 2012 at 05:25:01AM +0400, George Shuklin wrote: >>> I've understand that major and minor numbers for device is 8-bit size. >>> >>> I've don't understand, WHY? In 2012! >>> >>> (I've just have some discussion about 8-bit counter limitations in >>> context of cloud computing. In middle of 2012!!!!) >> Yes, shocking. It's been several years and you are still using that >> buzzword; I mean, it's _so_ 2010... Should've moved on to whatever's >> in this year... >> >> Incidentally, minors are 20 bit and majors are 12, but don't let that >> stand in the way of righteous indignation - what's mere facts when >> one is having discussions in context, presumably leveraging synergies >> all along... > Ok, thank you for information. > > But question is almost same: can I have 30k active logical volumes on > my single server, please? What are you trying to do, create a logical volume for every file? There are at least three separate technologies available that will solve whatever your problem is better than having 30,000 logical volumes. I don't have a clue what you think you're trying to accomplish and I am still willing to bet beers that you're bus is parked solidly on the wrong tracks. > -- > To unsubscribe from this list: send the line "unsubscribe > linux-kernel" in > the body of a message to majordomo@vger.kernel.org > More majordomo info at http://vger.kernel.org/majordomo-info.html > Please read the FAQ at http://www.tux.org/lkml/ > ^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: Why minor is still 8 bit? 2012-06-04 16:11 ` Casey Schaufler @ 2012-06-04 18:59 ` George Shuklin 2012-06-05 0:52 ` H. Peter Anvin 0 siblings, 1 reply; 11+ messages in thread From: George Shuklin @ 2012-06-04 18:59 UTC (permalink / raw) To: Casey Schaufler; +Cc: Al Viro, linux-kernel On 04.06.2012 20:11, Casey Schaufler wrote: > On 6/3/2012 8:53 PM, George Shuklin wrote: >> On 04.06.2012 05:35, Al Viro wrote: >>> On Mon, Jun 04, 2012 at 05:25:01AM +0400, George Shuklin wrote: >>>> I've understand that major and minor numbers for device is 8-bit size. >>>> >>>> I've don't understand, WHY? In 2012! >>>> >>>> (I've just have some discussion about 8-bit counter limitations in >>>> context of cloud computing. In middle of 2012!!!!) >>> Yes, shocking. It's been several years and you are still using that >>> buzzword; I mean, it's _so_ 2010... Should've moved on to whatever's >>> in this year... >>> >>> Incidentally, minors are 20 bit and majors are 12, but don't let that >>> stand in the way of righteous indignation - what's mere facts when >>> one is having discussions in context, presumably leveraging synergies >>> all along... >> Ok, thank you for information. >> >> But question is almost same: can I have 30k active logical volumes on >> my single server, please? > What are you trying to do, create a logical volume for every file? > There are at least three separate technologies available that will > solve whatever your problem is better than having 30,000 logical > volumes. I don't have a clue what you think you're trying to > accomplish and I am still willing to bet beers that you're bus is > parked solidly on the wrong tracks. > > Very simple sample: I'd like to create shared storage to publish volumes via ISCSI. ~60Tb of drives, ~2Gb average disk size = 30k disk images. Can I just create a bunch of LV and export them by iet or scst? Nope: There is a serious limit for amount of active LV per host. Yes, I can create filesystem, put images (as file) to that filesystem and publish them back, but why FS is needed to do such simple task? ^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: Why minor is still 8 bit? 2012-06-04 18:59 ` George Shuklin @ 2012-06-05 0:52 ` H. Peter Anvin 2012-06-05 1:08 ` George Shuklin 0 siblings, 1 reply; 11+ messages in thread From: H. Peter Anvin @ 2012-06-05 0:52 UTC (permalink / raw) To: George Shuklin; +Cc: Casey Schaufler, Al Viro, linux-kernel On 06/04/2012 11:59 AM, George Shuklin wrote: > Very simple sample: I'd like to create shared storage to publish volumes > via ISCSI. ~60Tb of drives, ~2Gb average disk size = 30k disk images. > Can I just create a bunch of LV and export them by iet or scst? Nope: > There is a serious limit for amount of active LV per host. Yes, I can > create filesystem, put images (as file) to that filesystem and publish > them back, but why FS is needed to do such simple task? You realize that with that many volumes, a logical volume manager *is* a filesystem, right? -hpa ^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: Why minor is still 8 bit? 2012-06-05 0:52 ` H. Peter Anvin @ 2012-06-05 1:08 ` George Shuklin 2012-06-05 9:23 ` Paul Bolle 0 siblings, 1 reply; 11+ messages in thread From: George Shuklin @ 2012-06-05 1:08 UTC (permalink / raw) To: H. Peter Anvin; +Cc: Casey Schaufler, Al Viro, linux-kernel On 05.06.2012 04:52, H. Peter Anvin wrote: > On 06/04/2012 11:59 AM, George Shuklin wrote: >> Very simple sample: I'd like to create shared storage to publish volumes >> via ISCSI. ~60Tb of drives, ~2Gb average disk size = 30k disk images. >> Can I just create a bunch of LV and export them by iet or scst? Nope: >> There is a serious limit for amount of active LV per host. Yes, I can >> create filesystem, put images (as file) to that filesystem and publish >> them back, but why FS is needed to do such simple task? > You realize that with that many volumes, a logical volume manager *is* a > filesystem, right? > > -hpa Well, I have nothing against filesystem with very small 'db-like' footprint: no directories, no attributes, extra-large allocation block, very fast initialization, no random file growth (very calm metadata without constant updating). If someone do have that type of FS (for disk images storage) - why not? Anyway, using LVM to provide logical volumes to customers seems be fine (at least by name). But that limitation for minor is breaking whole idea. I don't ask 'please fix', I'm just curious why so small? ^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: Why minor is still 8 bit? 2012-06-05 1:08 ` George Shuklin @ 2012-06-05 9:23 ` Paul Bolle 2012-06-05 9:36 ` George Shuklin 0 siblings, 1 reply; 11+ messages in thread From: Paul Bolle @ 2012-06-05 9:23 UTC (permalink / raw) To: George Shuklin; +Cc: H. Peter Anvin, Casey Schaufler, Al Viro, linux-kernel On Tue, 2012-06-05 at 05:08 +0400, George Shuklin wrote: > Anyway, using LVM to provide logical volumes to customers seems be fine > (at least by name). But that limitation for minor is breaking whole idea. man 8 vgcreate reads (in part): -l, --maxlogicalvolumes MaxLogicalVolumes Sets the maximum number of logical volumes allowed in this volume group. The setting can be changed with vgchange. For volume groups with metadata in lvm1 format, the limit and default value is 255. If the metadata uses lvm2 format, the default value is 0 which removes this restriction: there is then no limit. Doesn't that option work for you (assuming you're not stuck on lvm1 format)? If not, where do things go wrong? (Note that I never worry about this limit, since I never use more than a handful of logical volumes per volume group.) Paul Bolle ^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: Why minor is still 8 bit? 2012-06-05 9:23 ` Paul Bolle @ 2012-06-05 9:36 ` George Shuklin 2012-06-05 9:49 ` Paul Bolle 0 siblings, 1 reply; 11+ messages in thread From: George Shuklin @ 2012-06-05 9:36 UTC (permalink / raw) To: Paul Bolle; +Cc: H. Peter Anvin, Casey Schaufler, Al Viro, linux-kernel On 05.06.2012 13:23, Paul Bolle wrote: > On Tue, 2012-06-05 at 05:08 +0400, George Shuklin wrote: >> Anyway, using LVM to provide logical volumes to customers seems be fine >> (at least by name). But that limitation for minor is breaking whole idea. > man 8 vgcreate reads (in part): > -l, --maxlogicalvolumes MaxLogicalVolumes > Sets the maximum number of logical volumes allowed in this > volume group. The setting can be changed with vgchange. For > volume groups with metadata in lvm1 format, the limit and > default value is 255. If the metadata uses lvm2 format, the > default value is 0 which removes this restriction: there is then > no limit. > > Doesn't that option work for you (assuming you're not stuck on lvm1 > format)? If not, where do things go wrong? > > (Note that I never worry about this limit, since I never use more than a > handful of logical volumes per volume group.) > This is a limitaiton of LMV itself. I'm talking about making those LV active (visible in system). When you trying to activate too much LV, you will get 'device io ctl failed' message. ^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: Why minor is still 8 bit? 2012-06-05 9:36 ` George Shuklin @ 2012-06-05 9:49 ` Paul Bolle 2012-06-05 11:13 ` George Shuklin 0 siblings, 1 reply; 11+ messages in thread From: Paul Bolle @ 2012-06-05 9:49 UTC (permalink / raw) To: George Shuklin; +Cc: H. Peter Anvin, Casey Schaufler, Al Viro, linux-kernel On Tue, 2012-06-05 at 13:36 +0400, George Shuklin wrote: > This is a limitaiton of LMV itself. I'm talking about making those LV > active (visible in system). When you trying to activate too much LV, you > will get 'device io ctl failed' message. What is the exact text of that message? Where is it found (stderr, dmesg, ...)? Do you know what prints it (kernel, user space)? Do you run into this issue when trying to activate the 256th LV, or later, or earlier? Paul Bolle ^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: Why minor is still 8 bit? 2012-06-05 9:49 ` Paul Bolle @ 2012-06-05 11:13 ` George Shuklin 0 siblings, 0 replies; 11+ messages in thread From: George Shuklin @ 2012-06-05 11:13 UTC (permalink / raw) To: Paul Bolle; +Cc: H. Peter Anvin, Casey Schaufler, Al Viro, linux-kernel On 05.06.2012 13:49, Paul Bolle wrote: > On Tue, 2012-06-05 at 13:36 +0400, George Shuklin wrote: >> This is a limitaiton of LMV itself. I'm talking about making those LV >> active (visible in system). When you trying to activate too much LV, you >> will get 'device io ctl failed' message. > What is the exact text of that message? Where is it found (stderr, > dmesg, ...)? Do you know what prints it (kernel, user space)? Do you run > into this issue when trying to activate the 256th LV, or later, or > earlier? > > > Paul Bolle > Ooops, I was deeply wrong. Sorry, the problem was related to LVM behavior in 2.6.32-12-xs (XenServer kernel). ^ permalink raw reply [flat|nested] 11+ messages in thread
end of thread, other threads:[~2012-06-05 11:13 UTC | newest] Thread overview: 11+ messages (download: mbox.gz follow: Atom feed -- links below jump to the message on this page -- 2012-06-04 1:25 Why minor is still 8 bit? George Shuklin 2012-06-04 1:35 ` Al Viro 2012-06-04 3:53 ` George Shuklin 2012-06-04 16:11 ` Casey Schaufler 2012-06-04 18:59 ` George Shuklin 2012-06-05 0:52 ` H. Peter Anvin 2012-06-05 1:08 ` George Shuklin 2012-06-05 9:23 ` Paul Bolle 2012-06-05 9:36 ` George Shuklin 2012-06-05 9:49 ` Paul Bolle 2012-06-05 11:13 ` George Shuklin
This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox; as well as URLs for NNTP newsgroup(s).