The net_unittests have gotten so large that on low end devices (in particular, v1 chromecast devices) they cannot complete in a reasonable time (see also: crbug/619993; part of the issue appears to be excessive overhead per test-batch).

This is a tracking bug for splitting apart the net_unittests. Some targets that appear like sizeable chunks to potentially split out:

net_base_unittests (net/base/*)
net_cert_unittests (net/cert/*)
net_disk_cache_unittests (net/disk_cache_/*)
net_http_unittests (net/http/*)
net_proxy_unittests (net/proxy/*)
net_quic_unittests (net/quic/*)
net_socket_unittests (net/socket/*)
net_spdy_unittests (net/spdy/*)
net_url_request_unittests (net/url_request/*)

As a proxy for number of tests in each directory:
$ for dir in net/*/; do echo $(ack -hc '(TEST\(|TEST_F\(|TEST_P\()' $dir) $dir; done | sort -n -r
1462 net/quic/
1023 net/http/
691 net/spdy/
635 net/cert/
417 net/url_request/
417 net/socket/
415 net/disk_cache/
370 net/base/
342 net/websockets/
291 net/proxy/
232 net/dns/
217 net/tools/
166 net/third_party/
157 net/cookies/
115 net/ftp/
64 net/ssl/
59 net/filter/
49 net/nqe/
43 net/log/
38 net/der/
37 net/server/
24 net/test/
21 net/extras/
20 net/sdch/
20 net/android/
17 net/cert_net/
0 net/interfaces/
0 net/docs/
0 net/data/