Commit 35f00690 authored by Rebecca Dodd's avatar Rebecca Dodd

Copy edits

parent 36235b9c
Pipeline #73476331 passed with stages
in 24 minutes and 1 second
......@@ -19,7 +19,7 @@ Our self-managed customers often encounter issues related to performance, or the
The Support team is working on becoming more data driven by using three new tools designed to aggregate and summarize performance data for self-managed customers. A focus on data-driven decision-making improves the customer relationship and demonstrates our commitment to making performance a key feature of GitLab.
We'll look at three open-source tools created by GitLab Self-Managed Support. Strace-Parser is a general tool that could be of use to anyone, while JSON Stats and GitLabSOS are tailored to GitLab, but could be easily modified.
We'll look at three open source tools created by GitLab Self-Managed Support. Strace parser is a general tool that could be of use to anyone, while JSON Stats and GitLabSOS are tailored to GitLab, but could be easily modified.
## 1. [Strace parser](https://gitlab.com/gitlab-com/support/toolbox/strace-parser)
......@@ -27,7 +27,7 @@ We'll look at three open-source tools created by GitLab Self-Managed Support. S
Unlike [newer](http://man7.org/linux/man-pages/man1/perf.1.html) and [more powerful](https://github.com/iovisor/bpftrace) tracing tools, strace adds [significant overhead to a process](http://www.brendangregg.com/blog/2014-05-11/strace-wow-much-syscall.html). However, strace is generally available even on very old versions of Linux.
An strace of a single-threaded program is linear, but following the threads of execution quickly gets difficult when there are many processes being captured. At GitLab Support we are typically tracing [Unicorn](https://bogomips.org/unicorn/) workers or [Gitaly](https://gitlab.com/gitlab-org/gitaly), which are highly concurrent, resulting in hundreds of process IDs being traced and hundreds of thousands of lines out of output from traces only a few seconds long.
An strace of a single-threaded program is linear, but following the threads of execution quickly gets difficult when there are many processes being captured. At GitLab Support we are typically tracing [Unicorn](https://bogomips.org/unicorn/) workers or [Gitaly](https://gitlab.com/gitlab-org/gitaly), which are highly concurrent, resulting in hundreds of process IDs being traced and hundreds of thousands of lines of output from traces only a few seconds long.
Will built [strace parser](https://gitlab.com/gitlab-com/support/toolbox/strace-parser) for these types of use cases. Strace parser summarizes the most meaningful processing data delivered by an strace in a more accessible format, allowing users to find the critical section sections of the data quickly.
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment