Limit of hosts monitored by proxy

[igorsh]

When server send configuration data on proxy, server expand macroses to their values and then send to proxy configuration with that values.
Suddenly, proxy stop to collect items where was macroses. There is an error in server log:

 12372:20120320:114643.892 [Z3005] query failed: [0] PGRES_FATAL_ERROR:ERROR:  syntax error at end of input at character 2048
 [select t.hostmacroid,t.hostid,t.macro,t.value from hostmacro t where t.hostid in (10001,10002,10020,10083,10091,10100,10106,10109,10114,10118,10150,10151,10152,10153,10154,10155,10157,10158,10159,10160,10161,10162,10166,10167,10186,10218,10220,10228,10238,10242,10243,10244,10310,10311,10315,10344,10345,10397,10399,10402,10407,10408,10412,10413,10414,10416,10417,10418,10425,10428,10430,10436,10438,10439,10441,10442,10445,10447,10448,10449,10450,10467,10474,10479,10480,10482,10483,10485,10520,10521,10523,10524,10525,10526,10527,10529,10530,10533,10534,10535,10536,10537,10538,10539,10540,10541,10542,10543,10544,10545,10546,10583,10584,10585,10586,10590,10591,10592,10594,10595,10596,10603,10616,10617,10621,10622,10623,10624,10625,10626,10627,10628,10633,10634,10636,10637,10638,10640,10641,10642,10647,10648,10649,10650,10651,10652,10665,10666,10667,10683,10686,10687,10688,10689,10691,10692,10693,10694,10695,10760,10768,10769,10774,10775,10776,10777,10778,10779,10781,10795,10796,10798,10801,10802,10803,10804,10806,10807,10810,10812,10813,10815,10816,10817,10818,10819,10820,10822,10823,10824,10825,10826,10827,10828,10829,10830,10831,10832,10833,10837,10838,10846,10850,10852,10853,10854,10857,10868,10869,10872,10873,10880,10882,10883,10887,10888,10889,10890,10891,10892,10893,10903,10951,10952,10963,10970,10971,10972,10973,10976,10978,10979,10982,10983,10984,10985,10986,10987,10988,10989,10990,10991,10992,10996,10999,11002,11006,11007,11009,11030,11035,11036,11046,11051,11052,11054,11057,11058,11059,11060,11062,11063,11065,11066,11067,11068,11070,11071,11076,11080,11081,11082,11085,11086,11087,11088,11094,11096,11098,11099,11100,11101,11102,11103,11104,11105,11106,11107,11108,11109,11110,11112,11113,11114,11121,11123,11125,11126,11127,11128,11129,11132,11134,11135,11136,11137,11140,11142,11143,11146,11147,11149,11150,11152,11155,11156,11158,11162,11163,11165,11166,11168,11169,11175,11176,11177,11178,11184,11185,11186,11191,11192,11193,11194,11195,11197,11201,11202,11205,11206,11207,11213,11215,11216,11217,11219,11220) or]

As you can see, query was shrink to exactly 2048 characters and end of query was lost so syntax become broken. Normal query looks like this:
select t.hostmacroid,t.hostid,t.macro,t.value from hostmacro t where t.hostid in (10001,10002,10020,10083, whatever) ordered by t.hostmacroid;
Because query failed, proxy didn't receive value of macro and try gather data with wrong parameters (for example it use {$COMMUNITY} as SNMP community name when try to get SNMP data - I see this with tcpdump).

I have zabbix-server 1.8.7 with several proxies.
DB for server is postgresql 8.4 on dedicated database server, so zabbix-server connect to it by network.
On that proxy with which I have problem is zabbix-proxy 1.8.5 and same database server.
Problem exists only on proxy which monitor >300 hosts with ~1700 items. There are 110 text/spaces/brackets characters in query so it leaves 1937 characters for hostids which give us theoretical limit ~322 hosts for one proxy. And it is limit for number of hosts and their linked templates.

I didn't find issues or threads in this forum with such problem, so I'm not sure is it a bug or feature.
To fix a problem I descrease number of hosts, monitored by one proxy. Is it recommended workaround or bug that must be fixed?