Re: [smartos-discuss] How to install SmartOS when memory is bigger than hard disk?
I am still trying to find a better way, so I read the source of /smartdc/lib/smartos_prompt_config.sh, then find the following comment before it create the swap zone: We cannot allow the swap size to be less than the size of DRAM, lest we run into the availrmem double accounting issue for locked anonymous memory that is backed by in-memory swap (which will severely and artificially limit VM tenancy). We will therfore not create a swap device smaller than DRAM -- but we still allow for the configuration variable to account for actual consumed space by using it to set the refreservation on the swap volume if/when the specified size is smaller than DRAM. What is "availrmem double accounting issue"? Does it mean that it is not a good idea to install smartos on large memory small disk machine? Or I can just ignore it, and set my favorite size of swap? Best, Si-Qi > 在 2016年11月26日,20:16,Si-Qi Liu写道: > > Dear Ian, > > Thank you for your advice. It works! Though I really don't like this > literally get-hands-dirty workaround... > > I pull most memories to make the machine "normal". Then install SmartOS, no > problem. Then resize the > swap zone to 2G and poweroff. Then insert the memories back and boot again. A > new problem arises here: > the svc:/system/filesystem/smartdc:default service fails. I check the log, > the reason is the dump zone is too > small. It seems that the default size of this zone is 1/100 of the whole > disk, so it is set to be 1.43G. But when > the machine has 256G memory, a dump zone should be at least 5.xxG (I don't > know how this number comes). > So I resize the dump zone to fulfill the requirement, and reboot again. There > is no problem this time. > > Best wishes, Si-Qi > > >> 在 2016年11月25日,17:29,Ian Collins 写道: >> >> On 11/25/16 10:05 PM, Si-Qi Liu wrote: >>> Hi, >>> >>> I am trying to install SmartOS on my HP Z820 workstation, which has 256G >>> memory and 160G SSD hard disk. >>> >>> There is no problem when I setup the network, the zpool, the root password, >>> and so on. Finally, when I press 'y' and >>> 'Enter' to start to installation, it fail. The reason is quite simple: when >>> the installing script create the swap zone, there >>> is no enough disk space, so it just stop there, and give me a prompt. I >>> tried to create a small swap zone by myself, >>> but I didn't know what to do next. Simply reboot doesn't work. >> >> Have you tried pulling some of the memory just for the install? >> >>> BTW, for a machine with 256G memory, is there necessary to create a swap >>> zone? If so, how big should it be? >> >> Once you have installed, you can resize the swap volume. I'd say yes, and >> for this machine, keep it small. >> >> Cheers. >> >> -- >> Ian. >> > > --- smartos-discuss Archives: https://www.listbox.com/member/archive/184463/=now RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00 Modify Your Subscription: https://www.listbox.com/member/?member_id=25769125_secret=25769125-7688e9fb Powered by Listbox: http://www.listbox.com
AW: [smartos-discuss] Wrong disksize calculation
@Gjermund Gusland Thorsen shure, i stopped the VM first @Jorge Both pools have a HDD mirror with an SSD mirror als log-device But that isn't really the problem. I think the problem is how SmartOS calculates the size from the ZVOL to its own MiB size inside the vm-json-definition... I was able to move the KVM to the other Host after i changed the volsize, for example: ### set set volsize=29G zones/uuid-disk0 gave me a "size": 29696 in the KVMs json ### After that, vmadm sees an integer instead of a float / double and i was able to move the KVM. Von: Jorge SchrauwenGesendet: Dienstag, 29. November 2016 10:34 An: smartos-discuss@lists.smartos.org Cc: Gjermund Gusland Thorsen Betreff: Re: [smartos-discuss] Wrong disksize calculation I vaguely remember having this when the blocksize did not match 4k from a HDD pool to 16k for SSD pool. Regards Jorge On 2016-11-29 10:26, Gjermund Gusland Thorsen wrote: > Did you stop the vm first? > > vmadm stop 557ba0c0-b09c-678f-b1bc-a6bf2cc7b439; vmadm send > 557ba0c0-b09c-678f-b1bc-a6bf2cc7b439 | ssh r...@192.168.xx.xx vmadm > receive > > G > > On 29 Nov, 2016, at 10:08, Kilian Ries wrote: > >> Hello, >> >> yesterday i encountered the following problem: >> >> I wanted to move a KVM from one host to another via "vmadm send" >> command, but that ended up in the following error: >> >> ### >> vmadm send 557ba0c0-b09c-678f-b1bc-a6bf2cc7b439 | ssh >> r...@192.168.xx.xx vmadm receive >> Invalid value(s) for: disks.*.size >> This socket is closed. >> ### >> >> >> Reading the vmadm manual i figured out that >> >> * disks.*.size MUST be an integer >> * if you have a ZVOL, disks.*.size is depending on the size from the >> ZVOL and cannot be updated via vmadm >> >> So i think SmartOS does somewhere calculate wrong disksizes, because >> there should always be an integer but that isn't the case in reallity, >> for example: >> >> -> first, output of vmadm get UUID >> -> second, output of zfs get all zones/UUID-disk0 >> >> ### >> >> Example 1: >> >> vmadm get UUID >> >> "disks": [ >> { >> ... >> "size": 117187.5, >> ... >> } >> ] >> >> >> volsize 114G >> >> >> >> Example 2: >> >> "disks": [ >> { >> ... >> "size": 29296,875 >> ... >> } >> ] >> >> >> volsize 28,6G >> >> ### >> >> >> Why does SmartOS calculate wrong numbers here? These should all be >> integers ... >> >> Thanks >> Greets >> Kilian >> >> >> smartos-discuss | Archives | Modify Your Subscription > > --- smartos-discuss Archives: https://www.listbox.com/member/archive/184463/=now RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00 Modify Your Subscription: https://www.listbox.com/member/?member_id=25769125_secret=25769125-7688e9fb Powered by Listbox: http://www.listbox.com
Re: [smartos-discuss] Wrong disksize calculation
I vaguely remember having this when the blocksize did not match 4k from a HDD pool to 16k for SSD pool. Regards Jorge On 2016-11-29 10:26, Gjermund Gusland Thorsen wrote: Did you stop the vm first? vmadm stop 557ba0c0-b09c-678f-b1bc-a6bf2cc7b439; vmadm send 557ba0c0-b09c-678f-b1bc-a6bf2cc7b439 | ssh r...@192.168.xx.xx vmadm receive G On 29 Nov, 2016, at 10:08, Kilian Rieswrote: Hello, yesterday i encountered the following problem: I wanted to move a KVM from one host to another via "vmadm send" command, but that ended up in the following error: ### vmadm send 557ba0c0-b09c-678f-b1bc-a6bf2cc7b439 | ssh r...@192.168.xx.xx vmadm receive Invalid value(s) for: disks.*.size This socket is closed. ### Reading the vmadm manual i figured out that • disks.*.size MUST be an integer • if you have a ZVOL, disks.*.size is depending on the size from the ZVOL and cannot be updated via vmadm So i think SmartOS does somewhere calculate wrong disksizes, because there should always be an integer but that isn't the case in reallity, for example: -> first, output of vmadm get UUID -> second, output of zfs get all zones/UUID-disk0 ### Example 1: vmadm get UUID "disks": [ { ... "size": 117187.5, ... } ] volsize 114G Example 2: "disks": [ { ... "size": 29296,875 ... } ] volsize 28,6G ### Why does SmartOS calculate wrong numbers here? These should all be integers ... Thanks Greets Kilian smartos-discuss | Archives | Modify Your Subscription --- smartos-discuss Archives: https://www.listbox.com/member/archive/184463/=now RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00 Modify Your Subscription: https://www.listbox.com/member/?member_id=25769125_secret=25769125-7688e9fb Powered by Listbox: http://www.listbox.com
Re: [smartos-discuss] Wrong disksize calculation
Did you stop the vm first? vmadm stop 557ba0c0-b09c-678f-b1bc-a6bf2cc7b439; vmadm send 557ba0c0-b09c-678f-b1bc-a6bf2cc7b439 | ssh r...@192.168.xx.xx vmadm receive G On 29 Nov, 2016, at 10:08, Kilian Rieswrote: > Hello, > > yesterday i encountered the following problem: > > I wanted to move a KVM from one host to another via "vmadm send" command, but > that ended up in the following error: > > ### > vmadm send 557ba0c0-b09c-678f-b1bc-a6bf2cc7b439 | ssh r...@192.168.xx.xx > vmadm receive > Invalid value(s) for: disks.*.size > This socket is closed. > ### > > > Reading the vmadm manual i figured out that > > • disks.*.size MUST be an integer > • if you have a ZVOL, disks.*.size is depending on the size from the > ZVOL and cannot be updated via vmadm > > So i think SmartOS does somewhere calculate wrong disksizes, because there > should always be an integer but that isn't the case in reallity, for example: > > -> first, output of vmadm get UUID > -> second, output of zfs get all zones/UUID-disk0 > > ### > > Example 1: > > vmadm get UUID > > "disks": [ > { > ... > "size": 117187.5, > ... > } > ] > > > volsize 114G > > > > Example 2: > > "disks": [ > { > ... > "size": 29296,875 > ... > } > ] > > > volsize 28,6G > > ### > > > Why does SmartOS calculate wrong numbers here? These should all be integers > ... > > Thanks > Greets > Kilian > > > smartos-discuss | Archives | Modify Your Subscription signature.asc Description: Message signed with OpenPGP using GPGMail --- smartos-discuss Archives: https://www.listbox.com/member/archive/184463/=now RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00 Modify Your Subscription: https://www.listbox.com/member/?member_id=25769125_secret=25769125-7688e9fb Powered by Listbox: http://www.listbox.com
Re: [smartos-discuss] Power Management on Modern CPUs
Still fighting on with this. Tried several configurations with the SSDs now (mirrored cache, striped cache, mirrored logs striped cache, individual cache and mirror), all eventually end with the same host reboot with nothing logged or in the crash dump directory. One thing I have noticed is that when any log device is configured, the system reboots more often. This happens on two boxes of identical configuration. I am beginning to wonder if either; - Support for the onboard SATA controller isn’t great - We are generating too much IO for the SATA controller or the SSDs to keep up with Does anybody else have any systems running on a C612 chipset box and SATA drives? > On 6 Oct 2016, at 12:17, Adam Richmond-Gordon> wrote: > >> If the device dies while the system is running the log will fall back to >> being on the pool and any uncommitted entries are still in memory and >> won't be lost. > > This is pretty much what I’d had in mind. That said, this is a production box > and data loss isn’t something I really want to be dealing with. Removing the > cache device hasn’t changed performance at all, so I guess there’s enough > free RAM on the box to deal with the majority of the ARC. > > It’ll be interesting (now that the drives are mirrored) to see if the > crashing still occurs, or if the drive that appears to be timing out just > gets dropped from the pool. > >> On 6 Oct 2016, at 04:13, Paul B. Henson wrote: >> >> On Thu, Oct 06, 2016 at 03:42:36AM +0100, Adam Richmond-Gordon wrote: >> >>> Thank you for pointing that out - I had it in my head that a single >>> log device would be safe. >> >> There's only one failure mode (AFAIK) where a single log device will >> cause data loss; if your box crashes or has an unclean shutdown (say due >> to a power failure) while there are uncommitted entries on the log >> device, and the device fails before the system comes back online to >> process them. >> >> If the device dies while the system is running the log will fall back to >> being on the pool and any uncommitted entries are still in memory and >> won't be lost. If the device dies after a clean shutdown or poweroff >> there won't be any uncommitted entries on it and when the system comes >> up it will fail the device and again just fall back to an on-pool log. >> >> So while it's true that a single log device is non-redundant, it's a lot >> less "not safe" than say a non-redundant pool. You'd have to be pretty >> unlucky to actually have data loss from losing a non-redundant log >> device. Of course, depending on the importance of your data, that might >> not be a risk you want to take. But for a budget sensitive system, a >> single high-cost SSD for a log isn't an insane configuration if you >> think the odds of your system crashing/powering off dirty at the exact >> same time your log device dies are pretty low. >> > > --- smartos-discuss Archives: https://www.listbox.com/member/archive/184463/=now RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00 Modify Your Subscription: https://www.listbox.com/member/?member_id=25769125_secret=25769125-7688e9fb Powered by Listbox: http://www.listbox.com