[Author Prev][Author Next][Thread Prev][Thread Next][Author Index][Thread Index]
Re: [tor-bugs] #22706 [Core Tor/Tor]: Compress cached descriptors and consensus
#22706: Compress cached descriptors and consensus
---------------------------------------+-----------------------------------
Reporter: nickm | Owner:
Type: defect | Status: new
Priority: Medium | Milestone: Tor:
| 0.3.2.x-final
Component: Core Tor/Tor | Version:
Severity: Normal | Resolution:
Keywords: sponsor8-maybe disk-usage | Actual Points:
Parent ID: #7176 | Points: 4
Reviewer: nickm | Sponsor:
---------------------------------------+-----------------------------------
Changes (by nickm):
* status: needs_review => new
* keywords: sponsor8-maybe disk-usage review-group-18 => sponsor8-maybe
disk-usage
Comment:
Review:
* I think this presumes #22704.
* This patch seems to have the same issue with anonymous mmaps as I
noted in #22704: is it really saving RAM? I can believe it saves disk when
Tor isn't running.
* This patch should IMO be off-by-default, and I don't think it's safe
to turn it on for directory caches.
* The compression parts need to be rewritten to use the new compress.c
APIs.
* There are bigger memory savings available on clients: once we have the
(micro)descriptors parsed, we can just leave them on disk, parse them, and
not keep their text in memory at all. (After parsing we only use the text
to answer controller requests and directory requests.)
* We shouldn't add our own functions in the gz* namespace.
* Cache compression won't work very well: it only handles one
(micro)descriptor at a time.
If we decide to take this approach, we should start with a fresh patch,
using parts of the design here as needed.
--
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/22706#comment:2>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online
_______________________________________________
tor-bugs mailing list
tor-bugs@xxxxxxxxxxxxxxxxxxxx
https://lists.torproject.org/cgi-bin/mailman/listinfo/tor-bugs