没有合适的资源?快使用搜索试试~ 我知道了~
ceph故障诊断与排错
需积分: 22 13 下载量 25 浏览量
2018-07-25
09:34:51
上传
评论
收藏 480KB PDF 举报
温馨提示
试读
20页
CEPH存储峰会PPT,讲解软件定义存储CEPH的常见性能故障解决办法。
资源推荐
资源详情
资源评论
Common Support Issues
and
How to Troubleshoot Them
Michael Hackett and Vikhyat Umrao
PSMEs
Red Hat, Inc.
What are we going to cover today?
● Slow/Blocked Requests
○ What is a slow request?
○ Possible causes
○ Types of Slow Requests
○ Common Troubleshooting Techniques
● Flapping OSD's when RGW buckets have millions of objects
○ Where to start
○ Possible causes
○ Temporary solutions
○ Permanent Solutions
The Dreaded Slow Request!!!
● What is a Slow request?
○ When Ceph detects a request that has taken
too long to process it will get flagged as a
slow request.
● A Slow request will log against an OSD when it
has been unable to service the request in it’s
op_queue queue for 30 seconds or more (default).
○ Configurable via osd_op_complaint_time.
Default is 30 seconds.
● Will be accompanied by blocked requests.
● Changing the osd_op_complaint_time is not
recommended as can lead to false reporting
issues.
Possible Causes
With the understanding of what a slow request is, what can cause these to occur?
● Problems with underlying hardware such as disk drives, controllers, hosts
(kernel or configuration), racks or networking equipment
● System load
● Improper configurables set on OSD’s (op_threads set to high)
● Improper balance in the cluster leading to ‘hot’ OSD’s
● Cluster configuration issues (too many/not enough PG’s per OSD)
● Cluster under backfill/recovery
● Deep scrubbing
● Compaction or splitting is occurring on the OSD node.
剩余19页未读,继续阅读
资源评论
vincent0003301
- 粉丝: 0
- 资源: 8
上传资源 快速赚钱
- 我的内容管理 展开
- 我的资源 快来上传第一个资源
- 我的收益 登录查看自己的收益
- 我的积分 登录查看自己的积分
- 我的C币 登录后查看C币余额
- 我的收藏
- 我的下载
- 下载帮助
安全验证
文档复制为VIP权益,开通VIP直接复制
信息提交成功