On 31/03/14 17:03, Daniel Malon wrote: > Hi, > > afaik the Joyent images have still the custom kernel but will most likely > switch to stock kernels. > More information on the „Joyent optimized kernels“ is here: > http://linux.joyent.com/ Thanks for the reference. It's good to know where things are happening. The README is a bit sparse, however.
Checking the kernels on each instance I found that they are the same... SmartOS Image: # sha1sum /boot/vmlinuz-2.6.32-431.el6.x86_64 a7a245d687eff671ed5e5b2800f416b696f6715f /boot/vmlinuz-2.6.32-431.el6.x86_64 CentOS Install: sha1sum /boot/vmlinuz-2.6.32-431.el6.x86_64 a7a245d687eff671ed5e5b2800f416b696f6715f /boot/vmlinuz-2.6.32-431.el6.x86_64 I was surprised to see a difference in the initramfs but I haven't delved any deeper: SmartOS Image: # sha1sum /boot/initramfs-2.6.32-431.el6.x86_64.img.orig 26ac58d1f7869b37be5c55051837c3d5c60585f0 /boot/initramfs-2.6.32-431.el6.x86_64.img CentOS Install: # sha1sum /boot/initramfs-2.6.32-431.el6.x86_64.img f651ed194d4409b9a030254b2b5b13c989e14b8a /boot/initramfs-2.6.32-431.el6.x86_64.img > About „the one big partition“ thing: > It is the system partition. You should add a second data disk for all your > stuff (that will automatically get formatted and mounted at /data). > I guess none will be happy if you set a fixed disk size for the image … > because honestly everyone wants to have it a different size. Obviously I'm having problems coming across from the RHEL mentality. Typically one would mount several volumes and use LVM: Filesystem Size Used Avail Use% Mounted on /dev/mapper/vg-root 1008M 239M 719M 25% / tmpfs 499M 0 499M 0% /dev/shm /dev/vda1 248M 33M 203M 14% /boot /dev/mapper/vg-home 1008M 34M 924M 4% /home /dev/mapper/vg-tmp 504M 17M 462M 4% /tmp /dev/mapper/vg-usr 3.0G 477M 2.4G 17% /usr /dev/mapper/vg-var 1008M 104M 854M 11% /var Is there an adverse impact of using LVM inside the SmartOS KVM instance? I see that one can add more disks but I haven't tried this. Is it easier to manage volumes from the SmartOS side? Is there documentation for this? Possibly it would look something like this: Filesystem Size Used Avail Use% Mounted on /dev/vda1 1008M 239M 719M 25% / tmpfs 499M 0 499M 0% /dev/shm /dev/vdb1 248M 33M 203M 14% /boot /dev/vdc1 1008M 34M 924M 4% /home /dev/vdd1 504M 17M 462M 4% /tmp /dev/vde1 3.0G 477M 2.4G 17% /usr /dev/vdf1 1008M 104M 854M 11% /var > > If you really want to rely on only one disk and need that to be larger than > the 10G you have to repartition and then yes - it might be easier to just > install from an iso. > You will miss the metadata tools if you convert that VM into an image and > provision later (giving pubkeys via metadata for example). > > If you’re interested in the tooling around the image creation for CentOS you > might find information here: https://github.com/joyent/mi-centos This information about the tooling is very useful. Thanks. I now know a bit about DUO (I had no idea). I'm still not entirely sure what people mean when they say 'metadata tools'. Is that related to the "customer_metadata" which can be passed via the JSON file when I create the CentOS Custom KVM? > > > — > Daniel > > Am 31.03.2014 um 07:18 schrieb Tom Robinson <[email protected]>: > >> Hi, >> >> I've asked briefly on the #smartos IRC but thought I might get a broader >> response to my questions here. >> >> Does anyone have thoughts to share on running imgadm available images for >> CentOS vs custom installed >> instances of CentOS? Is there something in the SmartOS image kernels that is >> relevant or are these >> just stock CentOS kernels? Also, generally, what is the strategy for KVM >> 'updates'? E.g., When >> CentOS release updated kernels, is it OK to just update the KVMs in place >> (yum update)? >> >> One thing that I don't like on the joyant KVM images is the 'one big >> partition' partitioning >> strategy for what will be a 'production' server. I've gone to the trouble of >> re-partitioning >> centos-6 2.6.0 (df81f45e-8f9f-11e3-a819-93fab527c81e). Then I made my own >> image (clone) so I can >> redeploy it without the re-partitioning headache. Am I going down a dead-end >> path with this strategy? >> >> Clearly, it was lot easier installing a custom CentOS KVM (booted from an >> ISO) than re-partitioning >> the stock SmartOS image. I installed acpid on the custom guest so that a >> call to 'vmadm stop UUID' >> from the GZ shutdown the guest instance without trouble and otherwise it >> seemed to run OK (test was >> only to boot it and configure the network, and try startup/shutdown >> control). I am concerned, >> however, that there's some link to the hypervisor I'm missing with a custom >> image. >> >> All comments are welcome. >> >> Kind regards, >> Tom >> >> -- >> >> Tom Robinson >> IT Manager/System Administrator >> >> MoTeC Pty Ltd >> >> 121 Merrindale Drive >> Croydon South >> 3136 Victoria >> Australia >> >> T: +61 3 9761 5050 >> F: +61 3 9761 5051 >> E: [email protected] >> >> >> >> >> >> ------------------------------------------- >> smartos-discuss >> Archives: https://www.listbox.com/member/archive/184463/=now >> RSS Feed: https://www.listbox.com/member/archive/rss/184463/21955186-c7d24c67 >> Modify Your Subscription: https://www.listbox.com/member/?&; >> Powered by Listbox: http://www.listbox.com > > > ------------------------------------------- > smartos-discuss > Archives: https://www.listbox.com/member/archive/184463/=now > RSS Feed: https://www.listbox.com/member/archive/rss/184463/25908535-8e7ef2de > Modify Your Subscription: https://www.listbox.com/member/?&; > Powered by Listbox: http://www.listbox.com ------------------------------------------- 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&id_secret=25769125-7688e9fb Powered by Listbox: http://www.listbox.com
signature.asc
Description: OpenPGP digital signature
