1
<?xml version="1.0" encoding="utf-8"?>
2
<html xmlns="http://www.w3.org/1999/xhtml">
4
<title>Change log for Varnish 2.1</title>
5
<link rel="stylesheet" type="text/css" href="changes.css"/>
8
<h1>Change log for Varnish 2.1</h1>
9
<h2>Changes between 2.0.6 and 2.1.0</h2>
13
<p>The changelog in 2.1.0 included syntax errors, causing
14
the generated changelog to be empty.</p>
17
<p>The help text for <span class="code">default_grace</span> was wrongly
18
formatted and included a syntax error. This has now been fixed.</p>
21
<p><span class="code">varnishd</span> now closes the file descriptor used
22
to read the management secret file (from the <span class="code">-S</span>
26
<p>The child would previously try to close every valid file
27
descriptor, something which could cause problems if the file
28
descriptor ulimit was set too high. We now keep track of all
29
the file descriptors we open and only close up to that number.
33
<p>ESI was partially broken in 2.1.0 due to a bug in the
34
rollback of session workspace. This has been fixed.</p>
37
<p>Reject the <span class="code">auth</span>command rather than crash if
38
there is no <span class="code">-S</span> parameter given.</p>
41
<p>Align pointers in allocated objects. This will in theory
42
make Varnish a tiny bit faster at the expense of slightly more
46
<p>Ensure the master process process id is updated in the
47
shared memory log file after we go into the background.</p>
50
<p><span class="code">HEAD</span> requests would be converted to GET
51
requests too early, which affected <span class="code">pass</span>
52
and <span class="code">pipe</span>. This has been fixed.</p>
55
<p>Update the documentation to point out that the TTL is no
56
longer taken into account to decide whether an object is
60
<p>Add support for completely obliterating an object and all
61
variants of it. Currently, this has to be done using inline C.</p>
64
<p>Add experimental support for the <span class="code">Range</span>
65
header. This has to be enabled using the parameter
66
http_range_support.</p>
69
<p>The <span class="code">critbit</span> hasher could get into a deadlock
70
and had a race condition. Both those have now been fixed.</p>
76
<p><span class="code">varnishsizes</span>, which is
77
like <span class="code">varnishhost</span>, but for the length of objects,