From mboxrd@z Thu Jan 1 00:00:00 1970 From: Julian Anastasov Subject: Re: [*v4 PATCH 0/3] IPVS: Backup Adding Ipv6 and Persistence support Date: Tue, 16 Nov 2010 00:29:11 +0200 (EET) Message-ID: References: <201011151931.26789.hans.schillstrom@ericsson.com> Mime-Version: 1.0 Return-path: In-Reply-To: <201011151931.26789.hans.schillstrom@ericsson.com> Sender: lvs-devel-owner@vger.kernel.org List-ID: Content-Type: TEXT/PLAIN; charset="us-ascii"; format="flowed" Content-Transfer-Encoding: 7bit To: Hans Schillstrom Cc: Simon Horms , LVS-Devel , Daniel Lezcano Hello, On Mon, 15 Nov 2010, Hans Schillstrom wrote: > This patch series adds/(updates) the following functionality > in the synchronization between master and backup daemons. > > - IPv6 > - Persistence Engine > - Firewall marks transferred > - Time-outs transferred. > - Flag field increased to 32 bits. Great work! I don't see other fatal problems. May be you can finally fix some grammatical and space problems and we can wait next weeks for comments about the new message format from other IPVS users. We also talked about checking if cp->pe matches cp->dest->svc->pe in ip_vs_find_dest(). Or may be it is already handled by the recent change that compares p->pe with cp->pe before calling ct_match in ip_vs_ct_in_get? v4 PATCH 1/3 - still the case for pe_data_len=0 and pe_name_len!=0 is not handled properly as error because we now ignore pe_name silently v4 PATCH 2/3 ip_vs_sync_conn: - I now see that we can calculate the padding from previous message, so that we can skip sending padding after the last connection: pad = (4 - (int) curr_sb->head) & 3; then we should clear the data before message: p = curr_sb->head; curr_sb->head += pad; while (pad--) *(p++) = 0; s = (union ip_vs_sync_conn *) p; - Some checks are not needed, may be cp->pe_data_len is enough: -if (cp->pe_data_len && cp->dest->svc && cp->pe && cp->pe->name) +if (cp->pe_data_len) v4 PATCH 3/3 OK Regards -- Julian Anastasov