Hi,
what is the right way to deal with the new cr repository for centos 5 when you are running your own mirror. In particular the following statement I don't fully understand:
- Once 5.7 is released, the 5.6/cr/ repositories will be removed ( they will not be available on vault.centos.org )
Christian.
Hi,
On 08/16/2011 08:45 AM, Christian Anthon wrote:
what is the right way to deal with the new cr repository for centos 5 when you are running your own mirror. In particular the following statement I don't fully understand:
The reason why we only have baseurl pointing to mirror.centos.org is that we dont expect cr/ to be around for a long time. I think its a safe assumption that 5.7 should be ready for release in a few days time from here.
- Once 5.7 is released, the 5.6/cr/ repositories will be removed ( they
will not be available on vault.centos.org )
The CR only contains rpms from the next-release, so once that release is online we want to drop the CR/ completely and have everyone move over ( via a yum update ) to using the new rpms in 5.7/; it also means that there will be no continued updates to the 5.6/cr/ once 5.7 is around and online.
Let me know if any of that isnt very clear and I will try to explain in more details
- KB
On 08/16/2011 11:27 AM, Karanbir Singh wrote:
Hi,
On 08/16/2011 08:45 AM, Christian Anthon wrote:
what is the right way to deal with the new cr repository for centos 5 when you are running your own mirror. In particular the following statement I don't fully understand:
The reason why we only have baseurl pointing to mirror.centos.org is that we dont expect cr/ to be around for a long time. I think its a safe assumption that 5.7 should be ready for release in a few days time from here.
Is it really worth the hassle to add and remove the repo just to get the packages a few days early?
- Once 5.7 is released, the 5.6/cr/ repositories will be removed ( they
will not be available on vault.centos.org )
The CR only contains rpms from the next-release, so once that release is online we want to drop the CR/ completely and have everyone move over ( via a yum update ) to using the new rpms in 5.7/; it also means that there will be no continued updates to the 5.6/cr/ once 5.7 is around and online.
Let me know if any of that isnt very clear and I will try to explain in more details
What bugs me about this is the manual overhead for adding and removing the repo on the systems. I presume when TUV 5.8 gets released and I want to be up-to-date I have to add the CentOS cr repo again and then remove it when CentOS 5.8 is out?
Why not create a repo "5cr" that can be enabled once and never be removed? Once 5.7 is out you can simply clear that repo on the server side which essentially means users will pick up the packages from the 5.7 release repo. Then once the 5.8 updates come in you can put them into the 5cr repo again starting a new rolling release cycle.
That way people interested in the rolling release would only have to enable this only once on their systems and stay on this rolling release train until they deliberately deactivate the repo again.
Regards, Dennis
On 08/16/2011 11:30 AM, Dennis Jacobfeuerborn wrote:
Is it really worth the hassle to add and remove the repo just to get the packages a few days early?
you need to add it in, one time. You wont need to remove anything. As long as you do not edit that file it will be managed by centos-release-5-7 and centos-release-cr-5-7 etc.
Because, as step 1, we did not want to default everyone over into moving a machine's state unexpectedly : we created this one time manual step. Depending on how things go and how people feel we could potentially remove that manual step in the future.
The important thing here is that you have an option to either get onto or not get onto the cr/ release process, and you can opt out whenever you like by removing the file. But as long as you are 'in', you wont need to do anything manual from this point on.
And yes, I think its worth getting the security fix's at least out to as many people as possible, as quickly as possible. Segregating these security fixs, their build deps and their link chains is super complex and in many cases of dubious value in the EL release frame. Which is why we have the security, bugfix and enhancement updates rolled in here.
What bugs me about this is the manual overhead for adding and removing the repo on the systems. I presume when TUV 5.8 gets released and I want to be up-to-date I have to add the CentOS cr repo again and then remove it when CentOS 5.8 is out?
no, if you leave the centos-release-cr installed on your machine, you wont need to do anything. just keep the yum updates going
That way people interested in the rolling release would only have to enable this only once on their systems and stay on this rolling release train until they deliberately deactivate the repo again.
What we have now should be pretty much the same thing. You can make a decision based on site and system role policy, once you opt-in, you stay opted-in.
- KB
Karanbir Singh wrote on Tue, 16 Aug 2011 12:21:06 +0100:
You wont need to remove anything. As long as you do not edit that file it will be managed by centos-release-5-7 and centos-release-cr-5-7 etc.
Well, surely those people using their own mirrors will have to do something ...
Kai
On 08/17/2011 09:31 AM, Kai Schaetzl wrote:
You wont need to remove anything. As long as you do not edit that file it will be managed by centos-release-5-7 and centos-release-cr-5-7 etc.
Well, surely those people using their own mirrors will have to do something ...
Sure, but then if someone is going to hardwire urls to point at specific content, its reasonable to expect them to manage that content locally.
the /cr/ repo isnt going to go away, the content from inside it will, so depending on how you rehash the baseurl, leave the $releasever tag in, and things should be fine. As long as you mirror the entire repo structure from centos.org
- KB
On Wed, 17 Aug 2011, Karanbir Singh wrote:
On 08/17/2011 09:31 AM, Kai Schaetzl wrote:
You wont need to remove anything. As long as you do not edit that file it will be managed by centos-release-5-7 and centos-release-cr-5-7 etc.
Well, surely those people using their own mirrors will have to do something ...
Sure, but then if someone is going to hardwire urls to point at specific content, its reasonable to expect them to manage that content locally.
the /cr/ repo isnt going to go away, the content from inside it will, so depending on how you rehash the baseurl, leave the $releasever tag in, and things should be fine. As long as you mirror the entire repo structure from centos.org
So to be clear, all you are going to do is empty the cr repo and have an empty set of repodata in its place so that yum does not throw errors. Is this correct?
If this is true, is it your plan to NOT update the centos-release-cr-5* rpm with each point release?
Regards,
On Aug 17, 2011, at 12:21 PM, Karanbir Singh mail-lists@karan.org wrote:
On 08/17/2011 09:31 AM, Kai Schaetzl wrote:
You wont need to remove anything. As long as you do not edit that file it will be managed by centos-release-5-7 and centos-release-cr-5-7 etc.
Well, surely those people using their own mirrors will have to do something ...
Sure, but then if someone is going to hardwire urls to point at specific content, its reasonable to expect them to manage that content locally.
the /cr/ repo isnt going to go away, the content from inside it will, so depending on how you rehash the baseurl, leave the $releasever tag in, and things should be fine. As long as you mirror the entire repo structure from centos.org
I wonder if this wouldn't cause a lot more unnecessary download traffic from the mirrors as most people who replicate the tree will automatically pull in /cr even if they don't need/want it?
For real rolling releases I might have just created a /cr repo at the top level for each branch like someone else suggested, but instead of wiping it out between point releases just keep throwing the updated packages in there and createrepo it.
You would have dozens of versions of the same package in there after a while, but on the bright side on can 'yum downgrade' a package all the way back to 5.0. Of course you would probably purge packages older then X-Y revisions old to keep things sane.
Just my $.02
-Ross
I successfully upgraded desktops using the CR repo and 'yum update'.
However after installing the CR repo on 2 VPS's running C 5.6 and doing a reboot I got a load of Kernel messages, for the first time ever.
--------------------------
--------------------- Kernel Begin ------------------------
WARNING: Kernel Errors Present rtc_cmos: probe of rtc_cmos failed with error -16 ...: 1 Time(s)
1 Time(s): RCU-based detection of stalled CPUs is disabled. 1 Time(s): Verbose stalled-CPUs detection is disabled. 1 Time(s): 0: 0x00000001 -> 0x000000a0 1 Time(s): 0: 0x00000100 -> 0x00010000 1 Time(s): #1 [0001a7b000 - 0001a8c000] XEN PAGETABLES 1 Time(s): #10 [00019ed000 - 00019ee000] BOOTMEM 1 Time(s): #11 [00019ee000 - 00019ef000] BOOTMEM 1 Time(s): #12 [00019ef000 - 00019f0000] BOOTMEM 1 Time(s): #13 [00019f0000 - 00019f1000] BOOTMEM 1 Time(s): #14 [0002400000 - 0002780000] MEMMAP 0 1 Time(s): #15 [0001a8c000 - 0001aa4000] BOOTMEM 1 Time(s): #16 [0001aa4000 - 0001abc000] BOOTMEM 1 Time(s): #17 [00019f1000 - 00019f2000] BOOTMEM 1 Time(s): #18 [00019f2000 - 00019f3000] BOOTMEM 1 Time(s): #19 [00019f3000 - 00019f4000] BOOTMEM 1 Time(s): #2 [0001000000 - 00019d70c0] TEXT DATA BSS 1 Time(s): #20 [00019ec0c0 - 00019ec1a0] BOOTMEM 1 Time(s): #21 [00019ec1c0 - 00019ec228] BOOTMEM 1 Time(s): #22 [00019ec240 - 00019ec2a8] BOOTMEM 1 Time(s): #23 [00019ec2c0 - 00019ec328] BOOTMEM 1 Time(s): #24 [00019ec340 - 00019ec360] BOOTMEM 1 Time(s): #25 [00019ec380 - 00019ec3a0] BOOTMEM 1 Time(s): #26 [00019ec3c0 - 00019ec3e0] BOOTMEM 1 Time(s): #27 [0001abc000 - 0001ad9000] BOOTMEM 1 Time(s): #28 [00019ec400 - 00019ec408] BOOTMEM 1 Time(s): #29 [00019ec440 - 00019ec448] BOOTMEM 1 Time(s): #3 [00019f8000 - 0001a7b000] XEN START INFO 1 Time(s): #30 [00019ec480 - 00019ec484] BOOTMEM 1 Time(s): #31 [00019ec4c0 - 00019ec4c8] BOOTMEM 1 Time(s): #32 [00019ec500 - 00019ec600] BOOTMEM 1 Time(s): #33 [00019ec600 - 00019ec648] BOOTMEM 1 Time(s): #34 [00019ec680 - 00019ec6c8] BOOTMEM 1 Time(s): #35 [00019f4000 - 00019f6000] BOOTMEM 1 Time(s): #4 [0000001000 - 0000003000] TRAMPOLINE 1 Time(s): #5 [0000003000 - 0000007000] ACPI WAKEUP 1 Time(s): #6 [0000008000 - 0000076000] PGTABLE 1 Time(s): #7 [00019d70c0 - 00019eb0c0] NODE_DATA 1 Time(s): #8 [00019eb0c0 - 00019ec0c0] BOOTMEM 1 Time(s): #9 [0000007000 - 0000007030] BOOTMEM 1 Time(s): DMA 0x00000001 -> 0x00001000 1 Time(s): DMA32 0x00001000 -> 0x00100000 1 Time(s): Magic number: 1:252:3141 1 Time(s): NODE_DATA [00000000019d70c0 - 00000000019eb0bf] 1 Time(s): Normal empty 1 Time(s): Xen: 0000000000000000 - 00000000000a0000 (usable) 1 Time(s): Xen: 00000000000a0000 - 0000000000100000 (reserved) 1 Time(s): Xen: 0000000000100000 - 0000000010000000 (usable) 1 Time(s): ACPI in unprivileged domain disabled 1 Time(s): ACPI: Interpreter disabled. 1 Time(s): APIC: disable apic facility 1 Time(s): APIC: switched to apic NOOP 1 Time(s): Adding 524284k swap on /dev/xvda2. Priority:-1 extents:1 across:524284k SS 1 Time(s): Allocating PCI resources starting at 10000000 (gap: 10000000:f0000000) 1 Time(s): BIOS-provided physical RAM map: 1 Time(s): Block layer SCSI generic (bsg) driver version 0.4 loaded (major 253) 1 Time(s): Booting paravirtualized kernel on Xen 1 Time(s): Brought up 1 CPUs 1 Time(s): Built 1 zonelists in Node order, mobility grouping on. Total pages: 64543 1 Time(s): CONFIG_NF_CT_ACCT is deprecated and will be removed soon. Please use 1 Time(s): CPU: Unsupported number of siblings 4 1 Time(s): Calibrating delay loop (skipped), value calculated using timer frequency.. 4999.99 BogoMIPS (lpj=2499996) 1 Time(s): Checking aperture... 1 Time(s): Command line: root=/dev/xvda1 ro selinux=0 ro 1 Time(s): Console: colour dummy device 80x25 1 Time(s): DMI not present or invalid. 1 Time(s): Dentry cache hash table entries: 32768 (order: 6, 262144 bytes) 1 Time(s): Detected 2499.996 MHz processor. 1 Time(s): Dquot-cache hash table entries: 512 (order 0, 4096 bytes) 1 Time(s): EXT3-fs (xvda1): mounted filesystem with ordered data mode 1 Time(s): EXT3-fs (xvda1): using internal journal 1 Time(s): EXT3-fs: barriers not enabled 1 Time(s): Faking a node at 0000000000000000-0000000010000000 1 Time(s): Fixed MDIO Bus: probed 1 Time(s): Freeing SMP alternatives: 20k freed 1 Time(s): Freeing unused kernel memory: 892k freed 1 Time(s): Grant table initialized 1 Time(s): Hierarchical RCU implementation. 1 Time(s): HugeTLB registered 2 MB page size, pre-allocated 0 pages 1 Time(s): IMA: No TPM chip found, activating TPM-bypass! 1 Time(s): IP route cache hash table entries: 2048 (order: 2, 16384 bytes) 1 Time(s): Initalizing network drop monitor service 1 Time(s): Initialising Xen virtual ethernet driver. 1 Time(s): Initializing XFRM netlink socket 1 Time(s): Initializing cgroup subsys cpu 1 Time(s): Initializing cgroup subsys cpuacct 1 Time(s): Initializing cgroup subsys cpuset 1 Time(s): Initializing cgroup subsys devices 1 Time(s): Initializing cgroup subsys freezer 1 Time(s): Initializing cgroup subsys memory 1 Time(s): Initializing cgroup subsys net_cls 1 Time(s): Initializing cgroup subsys ns 1 Time(s): Initmem setup node 0 0000000000000000-0000000010000000 1 Time(s): Inode-cache hash table entries: 16384 (order: 5, 131072 bytes) 1 Time(s): Kernel command line: root=/dev/xvda1 ro selinux=0 ro 1 Time(s): Kernel log daemon terminating. 1 Time(s): Kernel logging (proc) stopped. 1 Time(s): Linux agpgart interface v0.103 1 Time(s): Linux version 2.6.35.4 (dfn@localhost.localdomain) (gcc version 4.4.4 20100630 (Red Hat 4.4.4-10) (GCC) ) #2 SMP Thu Sep 16 18:27:27 EDT 2010 1 Time(s): Memory: 246604k/262144k available (4193k kernel code, 388k absent, 15152k reserved, 3589k data, 892k init) 1 Time(s): Mount-cache hash table entries: 256 1 Time(s): Movable zone start PFN for each node 1 Time(s): NET: Registered protocol family 1 1 Time(s): NET: Registered protocol family 10 1 Time(s): NET: Registered protocol family 16 1 Time(s): NET: Registered protocol family 17 1 Time(s): NET: Registered protocol family 2 1 Time(s): NR_IRQS:16640 nr_irqs:256 1 Time(s): NX (Execute Disable) protection: active 1 Time(s): NetLabel: domain hash size = 128 1 Time(s): NetLabel: protocols = UNLABELED CIPSOv4 1 Time(s): NetLabel: unlabeled traffic allowed by default 1 Time(s): NetLabel: Initializing 2 Time(s): No AGP bridge found 1 Time(s): No NUMA configuration found 1 Time(s): No local APIC present 1 Time(s): Non-volatile memory driver v1.3 1 Time(s): PCI: Fatal: No config space access function found 2 Time(s): PCI: System does not support PCI 1 Time(s): PERCPU: Embedded 29 pages/cpu @ffff880001abc000 s90112 r8192 d20480 u118784 1 Time(s): PID hash table entries: 1024 (order: 1, 8192 bytes) 1 Time(s): PM: Registered nosave memory: 00000000000a0000 - 0000000000100000 1 Time(s): PNP: No PS/2 controller found. Probing ports directly. 1 Time(s): Performance Events: unsupported p6 CPU model 23 no PMU driver, software events only. 1 Time(s): Policy zone: DMA32 1 Time(s): SCSI subsystem initialized 1 Time(s): SELinux: Disabled at boot. 1 Time(s): SFI: Simple Firmware Interface v0.81 http://simplefirmware.org 1 Time(s): SLUB: Genslabs=14, HWalign=64, Order=0-3, MinObjects=0, CPUs=1, Nodes=1 1 Time(s): SMP alternatives: switching to UP code 1 Time(s): SMP: Allowing 1 CPUs, 0 hotplug CPUs 1 Time(s): Security Framework initialized 1 Time(s): Serial: 8250/16550 driver, 4 ports, IRQ sharing enabled 1 Time(s): Subtract (36 early reservations) 1 Time(s): Switching to clocksource xen 1 Time(s): TCP bind hash table entries: 8192 (order: 5, 131072 bytes) 1 Time(s): TCP cubic registered 1 Time(s): TCP established hash table entries: 8192 (order: 5, 131072 bytes) 1 Time(s): TCP reno registered 1 Time(s): TCP: Hash tables configured (established 8192 bind 8192) 1 Time(s): Time: 165:165:165 Date: 165/165/65 1 Time(s): UDP hash table entries: 128 (order: 0, 4096 bytes) 1 Time(s): UDP-Lite hash table entries: 128 (order: 0, 4096 bytes) 1 Time(s): VFS: Disk quotas dquot_6.5.2 1 Time(s): VFS: Mounted root (ext3 filesystem) readonly on device 202:1. 1 Time(s): XENBUS: Device with no driver: device/console/0 1 Time(s): Xen version: 3.3.0 (preserve-AD) 1 Time(s): Xen: using vcpu_info placement 1 Time(s): Zone PFN ranges: 1 Time(s): acpiphp: ACPI Hot Plug PCI Controller Driver version: 0.5 1 Time(s): alg: No test for stdrng (krng) 1 Time(s): allocated 2621440 bytes of page_cgroup 1 Time(s): audit: initializing netlink socket (disabled) 1 Time(s): bio: create slab <bio-0> at 0 1 Time(s): blkfront: xvda1: barriers enabled 1 Time(s): blkfront: xvda2: barriers enabled 1 Time(s): brd: module loaded 1 Time(s): console [hvc0] enabled 1 Time(s): console [tty0] enabled 1 Time(s): cpuidle: using governor ladder 1 Time(s): cpuidle: using governor menu 1 Time(s): device-mapper: ioctl: 4.17.0-ioctl (2010-03-05) initialised: dm-devel@redhat.com 1 Time(s): device-mapper: multipath: version 1.1.1 loaded 1 Time(s): device-mapper: uevent: version 1.0.3 1 Time(s): devtmpfs: initialized 1 Time(s): devtmpfs: mounted 1 Time(s): drivers/rtc/hctosys.c: unable to open rtc device (rtc0) 1 Time(s): early_node_map[2] active PFN ranges 1 Time(s): ftrace: allocating 19525 entries in 77 pages 1 Time(s): ftrace: converting mcount calls to 0f 1f 44 00 00 1 Time(s): i8042.c: No controller found. 1 Time(s): init_memory_mapping: 0000000000000000-0000000010000000 1 Time(s): input: PC Speaker as /devices/platform/pcspkr/input/input0 1 Time(s): installing Xen timer for CPU 0 1 Time(s): io scheduler cfq registered (default) 1 Time(s): io scheduler deadline registered 1 Time(s): io scheduler noop registered 1 Time(s): ip_tables: (C) 2000-2006 Netfilter Core Team 1 Time(s): kjournald starting. Commit interval 5 seconds 1 Time(s): klogd 1.4.1, log source = /proc/kmsg started. 1 Time(s): last_pfn = 0x10000 max_arch_pfn = 0x400000000 1 Time(s): lo: Disabled Privacy Extensions 2 Time(s): md: ... autorun DONE. 2 Time(s): md: Autodetecting RAID arrays. 1 Time(s): md: If you don't use raid, use raid=noautodetect 2 Time(s): md: Scanned 0 and added 0 devices. 1 Time(s): md: Waiting for all devices to be available before autodetect 2 Time(s): md: autorun ... 1 Time(s): mice: PS/2 mouse device common for all mice 1 Time(s): msgmni has been set to 481 1 Time(s): nf_conntrack version 0.5.0 (1926 buckets, 7704 max) 1 Time(s): nf_conntrack.acct=1 kernel parameter, acct=1 nf_conntrack module option or 1 Time(s): pci-stub: invalid id string "" 1 Time(s): pci_hotplug: PCI Hot Plug PCI Core version: 0.5 1 Time(s): pciehp: PCI Express Hot Plug Controller Driver version: 0.4 1 Time(s): pcpu-alloc: [0] 0 1 Time(s): pcpu-alloc: s90112 r8192 d20480 u118784 alloc=29*4096 1 Time(s): pid_max: default: 32768 minimum: 301 1 Time(s): platform rtc_cmos: registered platform RTC device (no PNP device found) 1 Time(s): please try 'cgroup_disable=memory' option if you don't want memory cgroups 1 Time(s): pnp: PnP ACPI: disabled 1 Time(s): registered taskstats version 1 1 Time(s): regulator: core version 0.5 1 Time(s): setup_percpu: NR_CPUS:256 nr_cpumask_bits:256 nr_cpu_ids:1 nr_node_ids:1 1 Time(s): sysctl net.netfilter.nf_conntrack_acct=1 to enable it. 1 Time(s): type=2000 audit(1313617098.849:1): initialized 1 Time(s): vgaarb: loaded 1 Time(s): warning: `dbus-daemon' uses 32-bit capabilities (legacy support in use) 1 Time(s): warning: process `kudzu' used the deprecated sysctl system call with 1.23. 1 Time(s): xen_balloon: Initialising balloon driver.
-
Always Learning -
Christian Anthon -
Dennis Jacobfeuerborn -
Kai Schaetzl -
Karanbir Singh -
me＠tdiehl.org -
Ross Walker