Ubuntu cloud instance hang at boot screen

When you reboot ubuntu cloud instance, sometimes it hang in boot screen with some lines such as:
random: nonblocking pool is initialized
or
random: init urandom read with 94 bits of entropy available

Reboot it, hard reboot but it was still stuck at that.

Solution: migrate instance to another compute host

Kernel hanging tty in CentOS 6.6

In CentOS 6.6, version kernel 2.6.32-504.el6.x86_64, sometimes I could not connect to the server.
I can run ssh but can not get tty console.
After googling, I saw a question in stackoverflow
http://stackoverflow.com/questions/26628274/kernel-hanging-the-tty-subsystem
/var/log/messages

 kernel: INFO: task bash:44739 blocked for more than 120 seconds.
 kernel:      Not tainted 2.6.32-504.el6.x86_64 #1
 kernel: "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
 kernel: bash          D 0000000000000006     0 44739      1 0x00000080
 kernel: ffff88023acaf898 0000000000000046 0000000000000000 ffff8803b5580d00
 kernel: ffff88023acaf868 ffffffff81271aa4 00004e9cd768ce30 ffff88043c279800
 kernel: ffff88043c2799e0 0000000105231542 ffff88023bbe5098 ffff88023acaffd8
 kernel: Call Trace:
 kernel: [] ? blk_queue_bio+0x494/0x610
 kernel: [] schedule_timeout+0x215/0x2e0
 kernel: [] wait_for_common+0x123/0x180
 kernel: [] ? default_wake_function+0x0/0x20
 kernel: [] wait_for_completion+0x1d/0x20
 kernel: [] flush_cpu_workqueue+0x61/0x90
 kernel: [] ? wq_barrier_func+0x0/0x20
 kernel: [] flush_workqueue+0x54/0x80
 kernel: [] flush_scheduled_work+0x15/0x20
 kernel: [] tty_ldisc_release+0x3c/0x90
 kernel: [] tty_release_dev+0x40b/0x5e0
 kernel: [] ? __dec_zone_page_state+0x2e/0x30
 kernel: [] tty_release+0x1e/0x30
 kernel: [] __fput+0xf5/0x210
 kernel: [] fput+0x25/0x30
 kernel: [] filp_close+0x5d/0x90
 kernel: [] put_files_struct+0x7f/0xf0
 kernel: [] exit_files+0x53/0x70
 kernel: [] do_exit+0x18d/0x870
 kernel: [] ? __sigqueue_free+0x3d/0x50
 kernel: [] ? __dequeue_signal+0x102/0x200
 kernel: [] do_group_exit+0x58/0xd0
 kernel: [] get_signal_to_deliver+0x1f6/0x460
 kernel: [] do_signal+0x75/0x800
 kernel: [] ? __audit_syscall_exit+0x25e/0x290
 kernel: [] do_notify_resume+0x90/0xc0
 kernel: [] int_signal+0x12/0x17

or

 kernel: INFO: task sshd:25489 blocked for more than 120 seconds.
 kernel:      Not tainted 2.6.32-504.el6.x86_64 #1
 kernel: "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
 kernel: sshd          D 0000000000000000     0 25489  26121 0x00000080
 kernel: ffff88023a8cf728 0000000000000082 0000000000000000 0000000000000000
 kernel: ffff88023a8cf7f8 ffffffff8105ca34 00006b83e91acd8b ffff88023a8cf708
 kernel: ffff88023a8cf880 00000001070847e8 ffff880239e48638 ffff88023a8cffd8
 kernel: Call Trace:
 kernel: [] ? find_busiest_group+0x244/0x9e0
 kernel: [] schedule_timeout+0x215/0x2e0
 kernel: [] wait_for_common+0x123/0x180
 kernel: [] ? default_wake_function+0x0/0x20
 kernel: [] wait_for_completion+0x1d/0x20
 kernel: [] flush_work+0x77/0xc0
 kernel: [] ? wq_barrier_func+0x0/0x20
 kernel: [] flush_delayed_work+0x54/0x70
 kernel: [] tty_flush_to_ldisc+0x15/0x20
 kernel: [] n_tty_poll+0x67/0x1d0
 kernel: [] tty_poll+0x8a/0xa0
 kernel: [] do_select+0x3c5/0x7c0
 kernel: [] ? ip_finish_output+0x148/0x310
 kernel: [] ? __pollwait+0x0/0xf0
 kernel: [] ? pollwake+0x0/0x60
 kernel: [] ? pollwake+0x0/0x60
 kernel: [] ? pollwake+0x0/0x60
 kernel: [] ? pollwake+0x0/0x60
 kernel: [] ? _spin_unlock_bh+0x1b/0x20
 kernel: [] ? release_sock+0xe5/0x110
 kernel: [] ? tcp_sendmsg+0x73c/0xa20
 kernel: [] ? sock_aio_write+0x19b/0x1c0
 kernel: [] ? tty_wakeup+0x3d/0x80
 kernel: [] core_sys_select+0x18a/0x2c0
 kernel: [] ? n_tty_read+0x3ad/0x950
 kernel: [] ? autoremove_wake_function+0x0/0x40
 kernel: [] sys_select+0x47/0x110

Solution: yum upgrade to install new kernel, and reboot system to using new kernel.

Fix SSH long delay connection in CentOS

Add 2 options in /etc/ssh/sshd_config

UseDNS no
GSSAPIAuthentication no

Reload: systemctl reload sshd 

Simulate fail disk in Linux

If you want to simulate fail disk when test any storage software such as CEPH, ScaleIO ...
Block any IO send to device:
echo offline > /sys/block/sda/device/state

Re-enable:

echo running > /sys/block/sda/device/state

Link:
https://access.redhat.com/documentation/en-US/Red_Hat_Enterprise_Linux/5/html/Online_Storage_Reconfiguration_Guide/removing_path-to-storage-device.html

Link hay về Flask

http://gouthamanbalaraman.com/blog/flask-app-directory-structure.html
https://realpython.com/blog/python/using-flask-login-for-user-management-with-flask/
http://blog.miguelgrinberg.com/post/using-celery-with-flask

Shift + Delete

Tôi vừa repair lại con PC chạy Windows 7 và xóa đi những thứ mà mình đã tích được từ rất lâu. Những thứ thường dùng nhưng lại không mang lại lợi ích gì cho tôi trong công việc và cuộc sống. 

Xóa đi trong cảm giác nhẹ tênh và thoải mái. 

Đón chờ những thách thức và cơ hội mới.

Openstack, QoS and CEPH RBD

Bài này được làm theo guide sau:
http://www.sebastien-han.fr/blog/2013/12/23/openstack-ceph-rbd-and-qos/

Có mấy thông số có thể áp cho volume:

• total_bytes_sec: the total allowed bandwidth for the guest per second
• read_bytes_sec: sequential read limitation (bytes/s)
• write_bytes_sec: sequential write limitation
• total_iops_sec: the total allowed IOPS for the guest per second
• read_iops_sec: random read limitation
• write_iops_sec: random write limitation

- Chuẩn bị 1 instance, tạo 1 disk default và attach vào instance.

- Tạo partion, tạo filesystem và mount vào sử dụng
- Test:
$ dd if=/dev/zero of=test.img bs=1M count=50 conv=fdatasync
Sử dụng fdatasync() để flush data xuống disk ( thay vì ghi tạm vào memorry buffer )
http://www.informit.com/articles/article.aspx?p=23618&seqNum=5
Kết quả: ~ 20MB/s

- Tạo QoS (limit write với tốc độ ~ 2MB/s)
# cinder qos-create low-write-bytes consumer="front-end" write_bytes_sec=2000000

- Tạo volume type
# cinder type-create low-write-bytes

- Gắn volume-type với qos
# cinder qos-associate <qos_specs> <volume_type_id>

- Tạo volume để test:
# cinder create --display-name vol_limit_speed --volume-type low-write-bytes 1

- Attach volume vào instance

- Test lại và kết quả. ~ 2MB/s